程序代写代做 algorithm C assembler assembly kernel compiler cache graph arm flex MIPS® Architecture for Programmers Volume II-A: The MIPS32® Instruction Set Manual

MIPS® Architecture for Programmers Volume II-A: The MIPS32® Instruction Set Manual
Document Number: MD00086 Revision 6.06 December 15, 2016

Public. This publication contains proprietary information which is subject to change without notice and is supplied ‘as is’, without any warranty of any kind.
The MIPS32® Instruction Set Manual, Revision 6.06

Table of Contents
Chapter 1: About This Book …………………………………………………………………………………………………… 2
1.1: Typographical Conventions …………………………………………………………………………………………………………. 3 1.1.1: Italic Text………………………………………………………………………………………………………………………….. 3 1.1.2: Bold Text………………………………………………………………………………………………………………………….. 3 1.1.3: Courier Text ……………………………………………………………………………………………………………………… 3
1.2: UNPREDICTABLE and UNDEFINED …………………………………………………………………………………………… 3 1.2.1: UNPREDICTABLE…………………………………………………………………………………………………………….. 3 1.2.2: UNDEFINED …………………………………………………………………………………………………………………….. 4 1.2.3: UNSTABLE ………………………………………………………………………………………………………………………. 4
1.3: Special Symbols in Pseudocode Notation……………………………………………………………………………………… 4 1.4: Notation for Register Field Accessibility ………………………………………………………………………………………… 7 1.5: For More Information ………………………………………………………………………………………………………………….. 9
Chapter 2: Guide to the Instruction Set …………………………………………………………………………………. 10
2.1: Understanding the Instruction Fields …………………………………………………………………………………………… 10 2.1.1: Instruction Fields ……………………………………………………………………………………………………………… 12 2.1.2: Instruction Descriptive Name and Mnemonic……………………………………………………………………….. 12 2.1.3: Format Field ……………………………………………………………………………………………………………………. 12 2.1.4: Purpose Field ………………………………………………………………………………………………………………….. 13 2.1.5: Description Field ……………………………………………………………………………………………………………… 13 2.1.6: Restrictions Field……………………………………………………………………………………………………………… 13 2.1.7: Availability and Compatibility Fields ……………………………………………………………………………………. 14 2.1.8: Operation Field………………………………………………………………………………………………………………… 15 2.1.9: Exceptions Field………………………………………………………………………………………………………………. 15 2.1.10: Programming Notes and Implementation Notes Fields………………………………………………………… 15
2.2: Operation Section Notation and Functions…………………………………………………………………………………… 16 2.2.1: Instruction Execution Ordering…………………………………………………………………………………………… 16 2.2.2: Pseudocode Functions……………………………………………………………………………………………………… 16
2.3: Op and Function Subfield Notation……………………………………………………………………………………………… 27 2.4: FPU Instructions ………………………………………………………………………………………………………………………. 27
Chapter 3: The MIPS32® Instruction Set ……………………………………………………………………………….. 29
3.1: Compliance and Subsetting……………………………………………………………………………………………………….. 29 3.1.1: Subsetting of Non-Privileged Architecture …………………………………………………………………………… 29 3.2: Alphabetical List of Instructions ………………………………………………………………………………………………….. 31 ABS.fmt ……………………………………………………………………………………………………………………………………… 32 ADD…………………………………………………………………………………………………………………………………………… 33 ADD.fmt……………………………………………………………………………………………………………………………………… 34 ADDI………………………………………………………………………………………………………………………………………….. 35 ADDIU ……………………………………………………………………………………………………………………………………….. 36 ADDIUPC …………………………………………………………………………………………………………………………………… 37 ADDU ………………………………………………………………………………………………………………………………………… 38 ALIGN………………………………………………………………………………………………………………………………………… 39 ALNV.PS ……………………………………………………………………………………………………………………………………. 41 ALUIPC ……………………………………………………………………………………………………………………………………… 43 AND…………………………………………………………………………………………………………………………………………… 44 ANDI………………………………………………………………………………………………………………………………………….. 45
3
The MIPS32® Instruction Set Manual, Revision 6.06

AUI ……………………………………………………………………………………………………………………………………………. 47 AUIPC ……………………………………………………………………………………………………………………………………….. 48 B ……………………………………………………………………………………………………………………………………………….. 49 BAL……………………………………………………………………………………………………………………………………………. 50 BALC …………………………………………………………………………………………………………………………………………. 52 BC …………………………………………………………………………………………………………………………………………….. 53 BC1EQZ BC1NEZ……………………………………………………………………………………………………………………….. 54 BC1F …………………………………………………………………………………………………………………………………………. 56 BC1FL ……………………………………………………………………………………………………………………………………….. 58 BC1T …………………………………………………………………………………………………………………………………………. 60 BC1TL ……………………………………………………………………………………………………………………………………….. 62 BC2EQZ BC2NEZ……………………………………………………………………………………………………………………….. 64 BC2F …………………………………………………………………………………………………………………………………………. 66 BC2FL ……………………………………………………………………………………………………………………………………….. 67 BC2T …………………………………………………………………………………………………………………………………………. 69 BC2TL ……………………………………………………………………………………………………………………………………….. 70 BEQ…………………………………………………………………………………………………………………………………………… 72 BEQL…………………………………………………………………………………………………………………………………………. 73 BGEZ…………………………………………………………………………………………………………………………………………. 75 BGEZAL …………………………………………………………………………………………………………………………………….. 76 B{LE,GE,GT,LT,EQ,NE}ZALC ………………………………………………………………………………………………………. 77 BGEZALL …………………………………………………………………………………………………………………………………… 80 BC ………………………………………………………………………………………………………………………………….. 82 BGEZL……………………………………………………………………………………………………………………………………….. 86 BGTZ…………………………………………………………………………………………………………………………………………. 88 BGTZL……………………………………………………………………………………………………………………………………….. 89 BITSWAP …………………………………………………………………………………………………………………………………..92 BLEZ …………………………………………………………………………………………………………………………………………. 94 BLEZL ……………………………………………………………………………………………………………………………………….. 95 BLTZ………………………………………………………………………………………………………………………………………….. 97 BLTZAL ……………………………………………………………………………………………………………………………………… 98 BLTZALL ……………………………………………………………………………………………………………………………………. 99 BLTZL………………………………………………………………………………………………………………………………………. 101 BNE …………………………………………………………………………………………………………………………………………. 103 BNEL ……………………………………………………………………………………………………………………………………….. 104 BOVC BNVC …………………………………………………………………………………………………………………………….. 107 BREAK …………………………………………………………………………………………………………………………………….. 109 C.cond.fmt ………………………………………………………………………………………………………………………………… 110 CACHE…………………………………………………………………………………………………………………………………….. 114 CACHEE ………………………………………………………………………………………………………………………………….. 121 CEIL.L.fmt ………………………………………………………………………………………………………………………………… 127 CEIL.W.fmt ……………………………………………………………………………………………………………………………….. 128 CFC1 ……………………………………………………………………………………………………………………………………….. 129 CFC2 ……………………………………………………………………………………………………………………………………….. 131 CLASS.fmt………………………………………………………………………………………………………………………………… 132 CLO …………………………………………………………………………………………………………………………………………. 134 CLZ………………………………………………………………………………………………………………………………………….. 135 CMP.condn.fmt………………………………………………………………………………………………………………………….. 136 COP2……………………………………………………………………………………………………………………………………….. 141 CRC32B, CRC32H, CRC32W……………………………………………………………………………………………………… 143 CRC32CB, CRC32CH, CRC32CW ………………………………………………………………………………………………. 146 CTC1 ……………………………………………………………………………………………………………………………………….. 149 CTC2 ……………………………………………………………………………………………………………………………………….. 152
The MIPS32® Instruction Set Manual, Revision 6.06 4

5
The MIPS32® Instruction Set Manual, Revision 6.06
CVT.D.fmt…………………………………………………………………………………………………………………………………. 153 CVT.L.fmt …………………………………………………………………………………………………………………………………. 154 CVT.PS.S…………………………………………………………………………………………………………………………………. 155 CVT.S.PL …………………………………………………………………………………………………………………………………. 157 CVT.S.PU…………………………………………………………………………………………………………………………………. 158 CVT.S.fmt…………………………………………………………………………………………………………………………………. 159 CVT.W.fmt………………………………………………………………………………………………………………………………… 160 DDIV ………………………………………………………………………………………………………………………………………… 161 DDIVU ……………………………………………………………………………………………………………………………………… 162 DERET …………………………………………………………………………………………………………………………………….. 163 DI…………………………………………………………………………………………………………………………………………….. 164 DIV ………………………………………………………………………………………………………………………………………….. 165 DIV MOD DIVU MODU ………………………………………………………………………………………………………………. 168 DIV.fmt …………………………………………………………………………………………………………………………………….. 170 DIVU………………………………………………………………………………………………………………………………………… 171 DVP …………………………………………………………………………………………………………………………………………. 172 EHB …………………………………………………………………………………………………………………………………………. 175 EI …………………………………………………………………………………………………………………………………………….. 176 ERET ……………………………………………………………………………………………………………………………………….. 177 ERETNC…………………………………………………………………………………………………………………………………… 179 EVP …………………………………………………………………………………………………………………………………………. 181 EXT …………………………………………………………………………………………………………………………………………. 183 FLOOR.L.fmt …………………………………………………………………………………………………………………………….. 185 FLOOR.W.fmt……………………………………………………………………………………………………………………………. 186 GINVI……………………………………………………………………………………………………………………………………….. 187 GINVT ……………………………………………………………………………………………………………………………………… 189 INS ………………………………………………………………………………………………………………………………………….. 192 J………………………………………………………………………………………………………………………………………………. 194 JAL ………………………………………………………………………………………………………………………………………….. 195 JALR………………………………………………………………………………………………………………………………………… 196 JALR.HB…………………………………………………………………………………………………………………………………… 198 JALX………………………………………………………………………………………………………………………………………… 201 JIALC……………………………………………………………………………………………………………………………………….. 203 JIC …………………………………………………………………………………………………………………………………………… 205 JR ……………………………………………………………………………………………………………………………………………. 206 JR.HB ………………………………………………………………………………………………………………………………………. 208 LB ……………………………………………………………………………………………………………………………………………. 211 LBE………………………………………………………………………………………………………………………………………….. 212 LBU …………………………………………………………………………………………………………………………………………. 213 LBUE ……………………………………………………………………………………………………………………………………….. 214 LDC1 ……………………………………………………………………………………………………………………………………….. 215 LDC2 ……………………………………………………………………………………………………………………………………….. 216 LDXC1……………………………………………………………………………………………………………………………………… 218 LH……………………………………………………………………………………………………………………………………………. 219 LHE …………………………………………………………………………………………………………………………………………. 220 LHU …………………………………………………………………………………………………………………………………………. 221 LHUE……………………………………………………………………………………………………………………………………….. 222 LL ……………………………………………………………………………………………………………………………………………. 223 LLE ………………………………………………………………………………………………………………………………………….. 225 LLWP……………………………………………………………………………………………………………………………………….. 228 LLWPE …………………………………………………………………………………………………………………………………….. 230 LSA …………………………………………………………………………………………………………………………………………. 232 LUI…………………………………………………………………………………………………………………………………………… 233

LUXC1……………………………………………………………………………………………………………………………………… 234 LW …………………………………………………………………………………………………………………………………………… 235 LWC1 ………………………………………………………………………………………………………………………………………. 236 LWC2 ………………………………………………………………………………………………………………………………………. 237 LWE…………………………………………………………………………………………………………………………………………. 239 LWL …………………………………………………………………………………………………………………………………………. 240 LWLE……………………………………………………………………………………………………………………………………….. 242 LWPC ………………………………………………………………………………………………………………………………………. 246 LWR ………………………………………………………………………………………………………………………………………… 247 LWRE ………………………………………………………………………………………………………………………………………. 251 LWXC1 …………………………………………………………………………………………………………………………………….. 254 MADD………………………………………………………………………………………………………………………………………. 255 MADD.fmt…………………………………………………………………………………………………………………………………. 256 MADDF.fmt MSUBF.fmt ……………………………………………………………………………………………………………… 259 MADDU ……………………………………………………………………………………………………………………………………. 261 MAX.fmt MIN.fmt MAXA.fmt MINA.fmt………………………………………………………………………………………….. 262 MFC0……………………………………………………………………………………………………………………………………….. 266 MFC1……………………………………………………………………………………………………………………………………….. 267 MFC2……………………………………………………………………………………………………………………………………….. 268 MFHC0 …………………………………………………………………………………………………………………………………….. 270 MFHC1 …………………………………………………………………………………………………………………………………….. 271 MFHC2 …………………………………………………………………………………………………………………………………….. 272 MFHI………………………………………………………………………………………………………………………………………… 273 MFLO ………………………………………………………………………………………………………………………………………. 274 MOV.fmt …………………………………………………………………………………………………………………………………… 275 MOVF ………………………………………………………………………………………………………………………………………. 276 MOVF.fmt …………………………………………………………………………………………………………………………………. 277 MOVN………………………………………………………………………………………………………………………………………. 279 MOVN.fmt…………………………………………………………………………………………………………………………………. 280 MOVT ………………………………………………………………………………………………………………………………………. 281 MOVT.fmt…………………………………………………………………………………………………………………………………. 282 MOVZ ………………………………………………………………………………………………………………………………………. 284 MOVZ.fmt …………………………………………………………………………………………………………………………………. 285 MSUB ………………………………………………………………………………………………………………………………………. 286 MSUB.fmt …………………………………………………………………………………………………………………………………. 287 MSUBU ……………………………………………………………………………………………………………………………………. 289 MTC0……………………………………………………………………………………………………………………………………….. 290 MTC1……………………………………………………………………………………………………………………………………….. 292 MTC2……………………………………………………………………………………………………………………………………….. 293 MTHC0 …………………………………………………………………………………………………………………………………….. 294 MTHC1 …………………………………………………………………………………………………………………………………….. 295 MTHC2 …………………………………………………………………………………………………………………………………….. 296 MTHI………………………………………………………………………………………………………………………………………… 297 MTLO ………………………………………………………………………………………………………………………………………. 298 MUL…………………………………………………………………………………………………………………………………………. 299 MUL MUH MULU MUHU ……………………………………………………………………………………………………………. 300 MUL.fmt……………………………………………………………………………………………………………………………………. 302 MULT……………………………………………………………………………………………………………………………………….. 303 MULTU…………………………………………………………………………………………………………………………………….. 304 NAL …………………………………………………………………………………………………………………………………………. 306 NEG.fmt……………………………………………………………………………………………………………………………………. 307 NMADD.fmt ………………………………………………………………………………………………………………………………. 308 NMSUB.fmt ………………………………………………………………………………………………………………………………. 310
The MIPS32® Instruction Set Manual, Revision 6.06 6

7
The MIPS32® Instruction Set Manual, Revision 6.06
NOP…………………………………………………………………………………………………………………………………………. 312 NOR ………………………………………………………………………………………………………………………………………… 313 OR …………………………………………………………………………………………………………………………………………… 314 ORI ………………………………………………………………………………………………………………………………………….. 315 PAUSE …………………………………………………………………………………………………………………………………….. 317 PLL.PS …………………………………………………………………………………………………………………………………….. 319 PLU.PS…………………………………………………………………………………………………………………………………….. 320 PREF ……………………………………………………………………………………………………………………………………….. 321 PREFE …………………………………………………………………………………………………………………………………….. 325 PREFX …………………………………………………………………………………………………………………………………….. 329 PUL.PS…………………………………………………………………………………………………………………………………….. 330 PUU.PS ……………………………………………………………………………………………………………………………………. 331 RDHWR……………………………………………………………………………………………………………………………………. 332 RDPGPR ………………………………………………………………………………………………………………………………….. 335 RECIP.fmt ………………………………………………………………………………………………………………………………… 336 RINT.fmt …………………………………………………………………………………………………………………………………… 337 ROTR ………………………………………………………………………………………………………………………………………. 339 ROTRV …………………………………………………………………………………………………………………………………….. 340 ROUND.L.fmt ……………………………………………………………………………………………………………………………. 341 ROUND.W.fmt…………………………………………………………………………………………………………………………… 342 RSQRT.fmt……………………………………………………………………………………………………………………………….. 343 SB……………………………………………………………………………………………………………………………………………. 344 SBE …………………………………………………………………………………………………………………………………………. 346 SC …………………………………………………………………………………………………………………………………………… 347 SCE …………………………………………………………………………………………………………………………………………. 351 SCWP………………………………………………………………………………………………………………………………………. 354 SCWPE ……………………………………………………………………………………………………………………………………. 358 SDBBP …………………………………………………………………………………………………………………………………….. 361 SDC1……………………………………………………………………………………………………………………………………….. 362 SDC2……………………………………………………………………………………………………………………………………….. 363 SDXC1 …………………………………………………………………………………………………………………………………….. 364 SEB …………………………………………………………………………………………………………………………………………. 365 SEH …………………………………………………………………………………………………………………………………………. 366 SEL.fmt…………………………………………………………………………………………………………………………………….. 368 SELEQZ SELNEZ ……………………………………………………………………………………………………………………… 369 SELEQZ.fmt SELNEQZ.fmt ………………………………………………………………………………………………………… 371 SH …………………………………………………………………………………………………………………………………………… 373 SHE …………………………………………………………………………………………………………………………………………. 374 SIGRIE …………………………………………………………………………………………………………………………………….. 375 SLL ………………………………………………………………………………………………………………………………………….. 376 SLLV………………………………………………………………………………………………………………………………………… 377 SLT………………………………………………………………………………………………………………………………………….. 378 SLTI…………………………………………………………………………………………………………………………………………. 379 SLTIU ………………………………………………………………………………………………………………………………………. 380 SLTU ……………………………………………………………………………………………………………………………………….. 381 SQRT.fmt …………………………………………………………………………………………………………………………………. 382 SRA …………………………………………………………………………………………………………………………………………. 383 SRAV……………………………………………………………………………………………………………………………………….. 384 SRL …………………………………………………………………………………………………………………………………………. 385 SRLV ……………………………………………………………………………………………………………………………………….. 386 SSNOP …………………………………………………………………………………………………………………………………….. 387 SUB …………………………………………………………………………………………………………………………………………. 388 SUB.fmt ……………………………………………………………………………………………………………………………………. 389

SUBU ………………………………………………………………………………………………………………………………………. 390 SUXC1 …………………………………………………………………………………………………………………………………….. 391 SW…………………………………………………………………………………………………………………………………………… 392 SWC1 ………………………………………………………………………………………………………………………………………. 393 SWC2 ………………………………………………………………………………………………………………………………………. 394 SWE ………………………………………………………………………………………………………………………………………… 396 SWL…………………………………………………………………………………………………………………………………………. 397 SWLE ………………………………………………………………………………………………………………………………………. 399 SWR ………………………………………………………………………………………………………………………………………… 401 SWRE………………………………………………………………………………………………………………………………………. 404 SWXC1…………………………………………………………………………………………………………………………………….. 406 SYNC ………………………………………………………………………………………………………………………………………. 407 SYNCI ……………………………………………………………………………………………………………………………………… 412 SYSCALL …………………………………………………………………………………………………………………………………. 415 TEQ …………………………………………………………………………………………………………………………………………. 416 TEQI ………………………………………………………………………………………………………………………………………… 417 TGE …………………………………………………………………………………………………………………………………………. 418 TGEI ………………………………………………………………………………………………………………………………………… 419 TGEIU ……………………………………………………………………………………………………………………………………… 420 TGEU ………………………………………………………………………………………………………………………………………. 421 TLBINV …………………………………………………………………………………………………………………………………….. 422 TLBINVF…………………………………………………………………………………………………………………………………… 425 TLBP ……………………………………………………………………………………………………………………………………….. 427 TLBR ……………………………………………………………………………………………………………………………………….. 428 TLBWI ……………………………………………………………………………………………………………………………………… 430 TLBWR…………………………………………………………………………………………………………………………………….. 432 TLT ………………………………………………………………………………………………………………………………………….. 434 TLTI…………………………………………………………………………………………………………………………………………. 435 TLTIU ………………………………………………………………………………………………………………………………………. 436 TLTU ……………………………………………………………………………………………………………………………………….. 437 TNE …………………………………………………………………………………………………………………………………………. 438 TNEI ………………………………………………………………………………………………………………………………………… 439 TRUNC.L.fmt…………………………………………………………………………………………………………………………….. 440 TRUNC.W.fmt …………………………………………………………………………………………………………………………… 441 WAIT ……………………………………………………………………………………………………………………………………….. 442 WRPGPR …………………………………………………………………………………………………………………………………. 444 WSBH………………………………………………………………………………………………………………………………………. 445 XOR…………………………………………………………………………………………………………………………………………. 446 XORI………………………………………………………………………………………………………………………………………… 447
Appendix A: Instruction Bit Encodings ……………………………………………………………………………….. 448
A.1: Instruction Encodings and Instruction Classes …………………………………………………………………………… 448 A.2: Instruction Bit Encoding Tables………………………………………………………………………………………………… 448 A.3: Floating Point Unit Instruction Format Encodings ……………………………………………………………………….. 459 A.4: Release 6 Instruction Encodings………………………………………………………………………………………………. 461
Appendix B: Revision History …………………………………………………………………………………………….. 466
The MIPS32® Instruction Set Manual, Revision 6.06 8

List of Figures
Figure 2.1: Example of Instruction Description …………………………………………………………………………………………. 11 Figure 2.2: Example of Instruction Fields…………………………………………………………………………………………………. 12 Figure 2.3: Example of Instruction Descriptive Name and Mnemonic ………………………………………………………….. 12 Figure 2.4: Example of Instruction Format ……………………………………………………………………………………………….. 12 Figure 2.5: Example of Instruction Purpose ……………………………………………………………………………………………… 13 Figure 2.6: Example of Instruction Description …………………………………………………………………………………………. 13 Figure 2.7: Example of Instruction Restrictions ………………………………………………………………………………………… 14 Figure 2.8: Example of Instruction Operation …………………………………………………………………………………………… 15 Figure 2.9: Example of Instruction Exception …………………………………………………………………………………………… 15 Figure 2.10: Example of Instruction Programming Notes …………………………………………………………………………… 16 Figure 2.11: COP_LW Pseudocode Function …………………………………………………………………………………………… 16 Figure 2.12: COP_LD Pseudocode Function……………………………………………………………………………………………. 17 Figure 2.13: COP_SW Pseudocode Function ………………………………………………………………………………………….. 17 Figure 2.14: COP_SD Pseudocode Function …………………………………………………………………………………………… 17 Figure 2.15: CoprocessorOperation Pseudocode Function ………………………………………………………………………… 18 Figure 2.16: MisalignedSupport Pseudocode Function ……………………………………………………………………………… 18 Figure 2.17: AddressTranslation Pseudocode Function …………………………………………………………………………….. 19 Figure 2.18: LoadMemory Pseudocode Function ……………………………………………………………………………………… 19 Figure 2.19: StoreMemory Pseudocode Function …………………………………………………………………………………….. 20 Figure 2.20: Prefetch Pseudocode Function…………………………………………………………………………………………….. 20 Figure 2.21: SyncOperation Pseudocode Function …………………………………………………………………………………… 21 Figure 2.22: ValueFPR Pseudocode Function………………………………………………………………………………………….. 21 Figure 2.23: StoreFPR Pseudocode Function ………………………………………………………………………………………….. 22 Figure 2.24: CheckFPException Pseudocode Function …………………………………………………………………………….. 23 Figure 2.25: FPConditionCode Pseudocode Function……………………………………………………………………………….. 23 Figure 2.26: SetFPConditionCode Pseudocode Function ………………………………………………………………………….. 24 Figure 2.27: sign_extend Pseudocode Functions ……………………………………………………………………………………… 24 Figure 2.28: memory_address Pseudocode Function ……………………………………………………………………………….. 25 Figure 2.29: Instruction Fetch Implicit memory_address Wrapping ……………………………………………………………… 25 Figure 2.30: AddressTranslation implicit memory_address Wrapping………………………………………………………….. 25 Figure 2.31: SignalException Pseudocode Function …………………………………………………………………………………. 26 Figure 2.32: SignalDebugBreakpointException Pseudocode Function ………………………………………………………… 26 Figure 2.33: SignalDebugModeBreakpointException Pseudocode Function…………………………………………………. 26 Figure 2.34: NullifyCurrentInstruction PseudoCode Function ……………………………………………………………………… 26 Figure 2.35: PolyMult Pseudocode Function ……………………………………………………………………………………………. 27 Figure 3.1: ALIGN operation (32-bit)……………………………………………………………………………………………………….. 39 Figure 3.2: Example of an ALNV.PS Operation………………………………………………………………………………………… 41 Figure 3.3: Usage of Address Fields to Select Index and Way………………………………………………………………….. 115 Figure 3.4: Usage of Address Fields to Select Index and Way………………………………………………………………….. 121 Figure 3.5: Operation of the EXT Instruction ………………………………………………………………………………………….. 183 Figure 3.6: Operation of the INS Instruction …………………………………………………………………………………………… 192 Figure 4.1: Unaligned Word Load Using LWL and LWR…………………………………………………………………………… 240 Figure 4.2: Bytes Loaded by LWL Instruction …………………………………………………………………………………………. 241 Figure 4.3: Unaligned Word Load Using LWLE and LWRE………………………………………………………………………. 242 Figure 4.4: Bytes Loaded by LWLE Instruction……………………………………………………………………………………….. 243 Figure 4.5: Unaligned Word Load Using LWL and LWR…………………………………………………………………………… 247 Figure 4.6: Bytes Loaded by LWR Instruction ………………………………………………………………………………………… 248
1 The MIPS32® Instruction Set Manual, Revision 6.06

Figure 4.7: Unaligned Word Load Using LWLE and LWRE………………………………………………………………………. 251 Figure 4.8: Bytes Loaded by LWRE Instruction ………………………………………………………………………………………. 252 Figure 5.9: Unaligned Word Store Using SWL and SWR …………………………………………………………………………. 397 Figure 5.10: Bytes Stored by an SWL Instruction ……………………………………………………………………………………. 398 Figure 5.11: Unaligned Word Store Using SWLE and SWRE …………………………………………………………………… 399 Figure 5.12: Bytes Stored by an SWLE Instruction………………………………………………………………………………….. 400 Figure 5.13: Unaligned Word Store Using SWR and SWL ……………………………………………………………………….. 401 Figure 5.14: Bytes Stored by SWR Instruction ……………………………………………………………………………………….. 402 Figure 5.15: Unaligned Word Store Using SWRE and SWLE …………………………………………………………………… 404 Figure 5.16: Bytes Stored by SWRE Instruction ……………………………………………………………………………………… 405 Figure A.1: Sample Bit Encoding Table …………………………………………………………………………………………………. 449
The MIPS32® Instruction Set Manual, Revision 6.06 2

List of Tables
Table 1.1: Symbols Used in Instruction Operation Statements……………………………………………………………………… 4 Table 1.2: Read/Write Register Field Notation …………………………………………………………………………………………… 7 Table 2.1: AccessLength Specifications for Loads/Stores………………………………………………………………………….. 20 Table 3.1: FPU Comparisons Without Special Operand Exceptions ………………………………………………………….. 111 Table 3.2: FPU Comparisons With Special Operand Exceptions for QNaNs ………………………………………………. 112 Table 3.3: Usage of Effective Address…………………………………………………………………………………………………… 114 Table 3.4: Encoding of Bits[17:16] of CACHE Instruction …………………………………………………………………………. 115 Table 3.5: Encoding of Bits [20:18] of the CACHE Instruction …………………………………………………………………… 116 Table 3.6: Usage of Effective Address…………………………………………………………………………………………………… 121 Table 3.7: Encoding of Bits[17:16] of CACHEE Instruction ………………………………………………………………………. 122 Table 3.8: Encoding of Bits [20:18] of the CACHEE Instruction ………………………………………………………………… 123 Table 3.1: Types of Global TLB Invalidates ……………………………………………………………………………………………. 189 Table 4.1: Special Cases for FP MAX, MIN, MAXA, MINA……………………………………………………………………….. 264 Table 5.2: Values of hint Field for PREF Instruction ………………………………………………………………………………… 322 Table 5.3: Values of hint Field for PREFE Instruction………………………………………………………………………………. 326 Table 5.4: RDHWR Register Numbers ………………………………………………………………………………………………….. 332 Table 5.5: Encodings of the Bits[10:6] of the SYNC instruction; the SType Field…………………………………………. 409 Table A.1: Symbols Used in the Instruction Encoding Tables …………………………………………………………………… 449 Table A.2: MIPS32 Encoding of the Opcode Field ………………………………………………………………………………….. 451 Table A.3: MIPS32 SPECIAL Opcode Encoding of Function Field ……………………………………………………………. 452 Table A.4: MIPS32 REGIMM Encoding of rt Field …………………………………………………………………………………… 452 Table A.5: MIPS32 SPECIAL2 Encoding of Function Field ………………………………………………………………………. 453 Table A.6: MIPS32 SPECIAL3 Encoding of Function Field for Release 2 of the Architecture………………………… 453 Table A.7: MIPS32 MOVCI6R Encoding of tf Bit …………………………………………………………………………………….. 453 Table A.8: MIPS32 SRL Encoding of Shift/Rotate …………………………………………………………………………………… 454 Table A.9: MIPS32 SRLV Encoding of Shift/Rotate…………………………………………………………………………………. 454 Table A.10: MIPS32 BSHFL Encoding of sa Field…………………………………………………………………………………… 454 Table A.11: MIPS32 COP0 Encoding of rs Field …………………………………………………………………………………….. 455 Table A.12: MIPS32 COP0 Encoding of Function Field When rs=CO………………………………………………………… 455 Table A.13: PCREL Encoding of Minor Opcode Field ……………………………………………………………………………… 455 Table A.14: MIPS32 Encoding of rs Field ………………………………………………………………………………………………. 456 Table A.15: MIPS32 COP1 Encoding of Function Field When rs=S…………………………………………………………… 456 Table A.16: MIPS32 COP1 Encoding of Function Field When rs=D ………………………………………………………….. 457 Table A.17: MIPS32 COP1 Encoding of Function Field When rs=W or L ………………………………………………….. 457 Table A.18: MIPS32 COP1 Encoding of Function Field When rs=PS ……………………………………………………….. 458 Table A.19: MIPS32 COP1 Encoding of tf Bit When rs=S, D, or PS6R, Function=MOVCF6R ………………………. 458 Table A.20: MIPS32 COP2 Encoding of rs Field …………………………………………………………………………………….. 458 Table A.21: MIPS32 COP1X6R Encoding of Function Field …………………………………………………………………….. 459 Table A.22: Floating Point Unit Instruction Format Encodings…………………………………………………………………… 459 Table A.23: Release 6 MUL/DIV encodings …………………………………………………………………………………………… 462 Table A.24: Release 6 PC-relative family encoding…………………………………………………………………………………. 462 Table A.25: Release 6 PC-relative family encoding bitstrings …………………………………………………………………… 463 Table A.26: B*C compact branch encodings ………………………………………………………………………………………….. 464
1 The MIPS32® Instruction Set Manual, Revision 6.06

Chapter 1
About This Book
The MIPS32® Instruction Set Manual comes as part of a multi-volume set.
• Volume I-A describes conventions used throughout the document set, and provides an introduction to the MIPS32® Architecture
• Volume I-B describes conventions used throughout the document set, and provides an introduction to the micro- MIPSTM Architecture
• Volume II-A provides detailed descriptions of each instruction in the MIPS32® instruction set
• Volume II-B provides detailed descriptions of each instruction in the microMIPS32TM instruction set
• Volume III describes the MIPS32® and microMIPS32TM Privileged Resource Architecture which defines and governs the behavior of the privileged resources included in a MIPS® processor implementation
• Volume IV-a describes the MIPS16eTM Application-Specific Extension to the MIPS32® Architecture. Beginning with Release 3 of the Architecture, microMIPS is the preferred solution for smaller code size. Release 6 removes MIPS16e: MIPS16e cannot be implemented with Release 6.
• Volume IV-b describes the MDMXTM Application-Specific Extension to the MIPS64® Architecture and microMIPS64TM. It is not applicable to the MIPS32® document set nor the microMIPS32TM document set. With Release 5 of the Architecture, MDMX is deprecated. MDMX and MSA can not be implemented at the same time. Release 6 removes MDMX: MDMX cannot be implemented with Release 6.
• Volume IV-c describes the MIPS-3D® Application-Specific Extension to the MIPS® Architecture. Release 6 removes MIPS-3D: MIPS-3D cannot be implemented with Release 6.
• Volume IV-d describes the SmartMIPS®Application-Specific Extension to the MIPS32® Architecture and the microMIPS32TM Architecture . Release 6 removes SmartMIPS: SmartMIPS cannot be implemented with Release 6, neither MIPS32 Release 6 nor MIPS64 Release 6.
• Volume IV-e describes the MIPS® DSP Module to the MIPS® Architecture.
• Volume IV-f describes the MIPS® MT Module to the MIPS® Architecture
• Volume IV-h describes the MIPS® MCU Application-Specific Extension to the MIPS® Architecture
• Volume IV-i describes the MIPS® Virtualization Module to the MIPS® Architecture
• Volume IV-j describes the MIPS® SIMD Architecture Module to the MIPS® Architecture
The MIPS32® Instruction Set Manual, Revision 6.06 2

About This Book
1.1 Typographical Conventions
This section describes the use of italic, bold and courier fonts in this book. 1.1.1 Italic Text
• is used for emphasis
• is used for bits, fields, and registers that are important from a software perspective (for instance, address bits used by software, and programmable fields and registers), and various floating point instruction formats, such as S and D
• is used for the memory access types, such as cached and uncached 1.1.2 Bold Text
• represents a term that is being defined
• is used for bits and fields that are important from a hardware perspective (for instance, register bits, which are
not programmable but accessible only to hardware)
• is used for ranges of numbers; the range is indicated by an ellipsis. For instance, 5..1 indicates numbers 5 through 1
• is used to emphasize UNPREDICTABLE and UNDEFINED behavior, as defined below. 1.1.3 Courier Text
Courier fixed-width font is used for text that is displayed on the screen, and for examples of code and instruction pseudocode.
1.2 UNPREDICTABLE and UNDEFINED
The terms UNPREDICTABLE and UNDEFINED are used throughout this book to describe the behavior of the pro- cessor in certain cases. UNDEFINED behavior or operations can occur only as the result of executing instructions in a privileged mode (i.e., in Kernel Mode or Debug Mode, or with the CP0 usable bit set in the Status register). Unpriv- ileged software can never cause UNDEFINED behavior or operations. Conversely, both privileged and unprivileged software can cause UNPREDICTABLE results or operations.
1.2.1 UNPREDICTABLE
UNPREDICTABLE results may vary from processor implementation to implementation, instruction to instruction, or as a function of time on the same implementation or instruction. Software can never depend on results that are UNPREDICTABLE. UNPREDICTABLE operations may cause a result to be generated or not. If a result is gener- ated, it is UNPREDICTABLE. UNPREDICTABLE operations may cause arbitrary exceptions.
UNPREDICTABLE results or operations have several implementation restrictions:
• Implementations of operations generating UNPREDICTABLE results must not depend on any data source
(memory or internal state) which is inaccessible in the current processor mode
3
The MIPS32® Instruction Set Manual, Revision 6.06

• UNPREDICTABLE operations must not read, write, or modify the contents of memory or internal state which is inaccessible in the current processor mode. For example, UNPREDICTABLE operations executed in user mode must not access memory or internal state that is only accessible in Kernel Mode or Debug Mode or in another process
• UNPREDICTABLE operations must not halt or hang the processor 1.2.2 UNDEFINED
UNDEFINED operations or behavior may vary from processor implementation to implementation, instruction to instruction, or as a function of time on the same implementation or instruction. UNDEFINED operations or behavior may vary from nothing to creating an environment in which execution can no longer continue. UNDEFINED opera- tions or behavior may cause data loss.
UNDEFINED operations or behavior has one implementation restriction:
• UNDEFINED operations or behavior must not cause the processor to hang (that is, enter a state from which there is no exit other than powering down the processor). The assertion of any of the reset signals must restore the processor to an operational state
1.2.3 UNSTABLE
UNSTABLE results or values may vary as a function of time on the same implementation or instruction. Unlike UNPREDICTABLE values, software may depend on the fact that a sampling of an UNSTABLE value results in a legal transient value that was correct at some point in time prior to the sampling.
UNSTABLE values have one implementation restriction:
• Implementations of operations generating UNSTABLE results must not depend on any data source (memory or
internal state) which is inaccessible in the current processor mode
1.3 Special Symbols in Pseudocode Notation
In this book, algorithmic descriptions of an operation are described using a high-level language pseudocode resem- bling Pascal. Special symbols used in the pseudocode notation are listed in Table 1.1.
Table 1.1 Symbols Used in Instruction Operation Statements
1.3 Special Symbols in Pseudocode Notation
Symbol
Meaning

Assignment
, ≠
Tests for equality and inequality

Bit string concatenation
xy
A y-bit string formed by y copies of the single-bit value x
b#n
A constant value n in base b. For instance 10#100 represents the decimal value 100, 2#100 represents the binary value 100 (decimal 4), and 16#100 represents the hexadecimal value 100 (decimal 256). If the “b#” prefix is omitted, the default base is 10.
0bn
A constant value n in base 2. For instance 0b100 represents the binary value 100 (decimal 4).
0xn
A constant value n in base 16. For instance 0x100 represents the hexadecimal value 100 (decimal 256).
The MIPS32® Instruction Set Manual, Revision 6.06 4

About This Book
Table 1.1 Symbols Used in Instruction Operation Statements (Continued)
Symbol
Meaning
xy z
Selection of bits y through z of bit string x. Little-endian bit notation (rightmost bit is 0) is used. If y is less than z, this expression is an empty (zero length) bit string.
x.bit[y]
Bit y of bitstring x. Alternative to the traditional MIPS notation xy.
x.bits[y..z]
Selection of bits y through z of bit string x. Alternative to the traditional MIPS notation xy z.
x.byte[y]
Byte y of bitstring x. Equivalent to the traditional MIPS notation x8*y+7 8*y.
x.bytes[y..z]
Selection of bytes y through z of bit string x. Alternative to the traditional MIPS notation x8*y+7 8*z
x halfword[y] x.word[i] x.doubleword[i]
Similar extraction of particular bitfields (used in e.g., MSA packed SIMD vectors).
x.bit31, x.byte0, etc.
Examples of abbreviated form of x.bit[y], etc. notation, when y is a constant.
x fieldy
Selection of a named subfield of bitstring x, typically a register or instruction encoding.
More formally described as “Field y of register x”.
For example, FIR.D = “the D bit of the Coprocessor 1 Floating-point Implementation Register (FIR)”.
, 
2’s complement or floating point arithmetic: addition, subtraction
*, 
2’s complement or floating point multiplication (both used for either)
div
2’s complement integer division
mod
2’s complement modulo

Floating point division

2’s complement less-than comparison

2’s complement greater-than comparison

2’s complement less-than or equal comparison

2’s complement greater-than or equal comparison
nor
Bitwise logical NOR
xor
Bitwise logical XOR
and
Bitwise logical AND
or
Bitwise logical OR
not
Bitwise inversion
&&
Logical (non-Bitwise) AND
<< Logical Shift left (shift in zeros at right-hand-side) >>
Logical Shift right (shift in zeros at left-hand-side)
GPRLEN
The length in bits (32 or 64) of the CPU general-purpose registers
GPR[x]
CPU general-purpose register x. The content of GPR[0] is always zero. In Release 2 of the Architecture, GPR[x] is a short-hand notation for SGPR[ SRSCtlCSS, x].
SGPR[s,x]
In Release 2 of the Architecture and subsequent releases, multiple copies of the CPU general-purpose regis- ters may be implemented. SGPR[s,x] refers to GPR set s, register x.
FPR[x]
Floating Point operand register x
FCC[CC]
Floating Point condition code CC. FCC[0] has the same value as COC[1]. Release 6 removes the floating point condition codes.
FPR[x]
Floating Point (Coprocessor unit 1), general register x
5 The MIPS32® Instruction Set Manual, Revision 6.06

1.3 Special Symbols in Pseudocode Notation Table 1.1 Symbols Used in Instruction Operation Statements (Continued)
Symbol
Meaning
CPR[z,x,s]
Coprocessor unit z, general register x, select s
CP2CPR[x]
Coprocessor unit 2, general register x
CCR[z,x]
Coprocessor unit z, control register x
CP2CCR[x]
Coprocessor unit 2, control register x
COC[z]
Coprocessor unit z condition signal
Xlat[x]
Translation of the MIPS16e GPR number x into the corresponding 32-bit GPR number
BigEndianMem
Endian mode as configured at chip reset (0  Little-Endian, 1  Big-Endian). Specifies the endianness of the memory interface (see LoadMemory and StoreMemory pseudocode function descriptions) and the endi- anness of Kernel and Supervisor mode execution.
BigEndianCPU
The endianness for load and store instructions (0  Little-Endian, 1  Big-Endian). In User mode, this endianness may be switched by setting the RE bit in the Status register. Thus, BigEndianCPU may be com- puted as (BigEndianMem XOR ReverseEndian).
ReverseEndian
Signal to reverse the endianness of load and store instructions. This feature is available in User mode only, and is implemented by setting the RE bit of the Status register. Thus, ReverseEndian may be computed as (SRRE and User mode).
LLbit
Bit of virtual state used to specify operation for instructions that provide atomic read-modify-write. LLbit is set when a linked load occurs and is tested by the conditional store. It is cleared, during other CPU operation, when a store to the location would no longer be atomic. In particular, it is cleared by exception return instruc- tions.
I:, I+n:, I-n:
This occurs as a prefix to Operation description lines and functions as a label. It indicates the instruction time during which the pseudocode appears to “execute.” Unless otherwise indicated, all effects of the current instruction appear to occur during the instruction time of the current instruction. No label is equivalent to a time label of I. Sometimes effects of an instruction appear to occur either earlier or later — that is, during the instruction time of another instruction. When this happens, the instruction operation is written in sections labeled with the instruction time, relative to the current instruction I, in which the effect of that pseudocode appears to occur. For example, an instruction may have a result that is not available until after the next instruction. Such an instruction has the portion of the instruction operation description that writes the result register in a section labeled I+1.
The effect of pseudocode statements for the current instruction labeled I+1 appears to occur “at the same time” as the effect of pseudocode statements labeled I for the following instruction. Within one pseudocode sequence, the effects of the statements take place in order. However, between sequences of statements for different instructions that occur “at the same time,” there is no defined order. Programs must not depend on a particular order of evaluation between such sections.
PC
The Program Counter value. During the instruction time of an instruction, this is the address of the instruc- tion word. The address of the instruction that occurs during the next instruction time is determined by assign- ing a value to PC during an instruction time. If no value is assigned to PC during an instruction time by any pseudocode statement, it is automatically incremented by either 2 (in the case of a 16-bit MIPS16e instruc- tion) or 4 before the next instruction time. A taken branch assigns the target address to the PC during the instruction time of the instruction in the branch delay slot.
In the MIPS Architecture, the PC value is only visible indirectly, such as when the processor stores the restart address into a GPR on a jump-and-link or branch-and-link instruction, or into a Coprocessor 0 register on an exception. Release 6 adds PC-relative address computation and load instructions. The PC value contains a full 32-bit address, all of which are significant during a memory reference.
The MIPS32® Instruction Set Manual, Revision 6.06 6

About This Book
Table 1.1 Symbols Used in Instruction Operation Statements (Continued)
Symbol
Meaning
ISA Mode
In processors that implement the MIPS16e Application Specific Extension or the microMIPS base architec- tures, the ISA Mode is a single-bit register that determines in which mode the processor is executing, as fol- lows:
In the MIPS Architecture, the ISA Mode value is only visible indirectly, such as when the processor stores a combined value of the upper bits of PC and the ISA Mode into a GPR on a jump-and-link or branch-and-link instruction, or into a Coprocessor 0 register on an exception.
Encoding
Meaning
0
The processor is executing 32-bit MIPS instructions
1
The processor is executing MIIPS16e or microMIPS instructions
PABITS
The number of physical address bits implemented is represented by the symbol PABITS. As such, if 36 phys- ical address bits were implemented, the size of the physical address space would be 2PABITS = 236 bytes.
FP32RegistersMode
Indicates whether the FPU has 32-bit or 64-bit floating point registers (FPRs). In MIPS32 Release 1, the FPU has 32, 32-bit FPRs, in which 64-bit data types are stored in even-odd pairs of FPRs. In MIPS64, (and optionally in MIPS32 Release2 and Release 3) the FPU has 32 64-bit FPRs in which 64-bit data types are stored in any FPR.
In MIPS32 Release 1 implementations, FP32RegistersMode is always a 0. MIPS64 implementations have a compatibility mode in which the processor references the FPRs as if it were a MIPS32 implementation. In such a case FP32RegisterMode is computed from the FR bit in the Status register. If this bit is a 0, the pro- cessor operates as if it had 32, 32-bit FPRs. If this bit is a 1, the processor operates with 32 64-bit FPRs.
The value of FP32RegistersMode is computed from the FR bit in the Status register.
InstructionInBranchDe- laySlot
Indicates whether the instruction at the Program Counter address was executed in the delay slot of a branch or jump. This condition reflects the dynamic state of the instruction, not the static state. That is, the value is false if a branch or jump occurs to an instruction whose PC immediately follows a branch or jump, but which is not executed in the delay slot of a branch or jump.
SignalException(excep- tion, argument)
Causes an exception to be signaled, using the exception parameter as the type of exception and the argument parameter as an exception-specific argument). Control does not return from this pseudocode function—the exception is signaled at the point of the call.
1.4 Notation for Register Field Accessibility
In this document, the read/write properties of register fields use the notations shown in Table 1.1.
Table 1.2 Read/Write Register Field Notation
Read/Write Notation
Hardware Interpretation
Software Interpretation
R/W
A field in which all bits are readable and writable by software and, potentially, by hardware.
Hardware updates of this field are visible by software read. Software updates of this field are visible by hardware read.
If the Reset State of this field is ‘‘Undefined’’, either software or hardware must initialize the value before the first read will return a predictable value. This should not be confused with the formal definition of UNDEFINED behavior.
7 The MIPS32® Instruction Set Manual, Revision 6.06

1.4 Notation for Register Field Accessibility Table 1.2 Read/Write Register Field Notation (Continued)
Read/Write Notation
Hardware Interpretation
Software Interpretation
R
A field which is either static or is updated only by hardware.
If the Reset State of this field is either ‘‘0’’, ‘‘Pre- set’’, or ‘‘Externally Set’’, hardware initializes this field to zero or to the appropriate state, respectively, on powerup. The term ‘‘Preset’’ is used to suggest that the processor establishes the appropriate state, whereas the term ‘‘Externally Set’’ is used to sug- gest that the state is established via an external source (e.g., personality pins or initialization bit stream). These terms are suggestions only, and are not intended to act as a requirement on the imple- mentation.
If the Reset State of this field is ‘‘Undefined’’, hard- ware updates this field only under those conditions specified in the description of the field.
A field to which the value written by software is ignored by hardware. Software may write any value to this field without affecting hardware behavior. Software reads of this field return the last value updated by hardware.
If the Reset State of this field is ‘‘Undefined’’, soft- ware reads of this field result in an UNPREDICT- ABLE value except after a hardware update done under the conditions specified in the description of the field.
R0
R0 = reserved, read as zero, ignore writes by soft- ware.
Hardware ignores software writes to an R0 field. Neither the occurrence of such writes, nor the val- ues written, affects hardware behavior.
Hardware always returns 0 to software reads of R0 fields.
The Reset State of an R0 field must always be 0.
If software performs an mtc0 instruction which writes a non-zero value to an R0 field, the write to the R0 field will be ignored, but permitted writes to other fields in the register will not be affected.
Architectural Compatibility: R0 fields are reserved, and may be used for not-yet-defined purposes in future revisions of the architecture.
When writing an R0 field, current software should only write either all 0s, or, preferably, write back the same value that was read from the field.
Current software should not assume that the value read from R0 fields is zero, because this may not be true on future hardware.
Future revisions of the architecture may redefine an R0 field, but must do so in such a way that software which is unaware of the new definition and either writes zeros or writes back the value it has read from the field will continue to work correctly.
Writing back the same value that was read is guaran- teed to have no unexpected effects on current or future hardware behavior. (Except for non-atomicity of such read-writes.)
Writing zeros to an R0 field may not be preferred because in the future this may interfere with the oper- ation of other software which has been updated for the new field definition.
The MIPS32® Instruction Set Manual, Revision 6.06 8

About This Book
Table 1.2 Read/Write Register Field Notation (Continued)
Read/Write Notation
Hardware Interpretation
Software Interpretation
0
Release 6
Release 6 legacy “0” behaves like R0 – read as zero, nonzero writes ignored.
Legacy “0” should not be defined for any new control register fields; R0 should be used instead.
HW returns 0 when read. Only zero should be written, or, value read from reg- HW ignores writes. ister.
pre-Release 6
pre-Release 6 legacy “0” – read as zero, nonzero writes UNDEFINED
A field which hardware does not update, and for which hardware can assume a zero value.
A field to which the value written by software must be zero. Software writes of non-zero values to this field may result in UNDEFINED behavior of the hardware. Software reads of this field return zero as long as all previous software writes are zero.
If the Reset State of this field is ‘‘Undefined’’, soft- ware must write this field with zero before it is guar- anteed to read as zero.
R/W0
Like R/W, except that writes of non-zero to a R/W0 field are ignored. E.g. Status.NMI
Hardware may set or clear an R/W0 bit.
Hardware ignores software writes of nonzero to an R/W0 field. Neither the occurrence of such writes, nor the values written, affects hardware behavior.
Software writes of 0 to an R/W0 field may have an effect.
Hardware may return 0 or nonzero to software reads of an R/W0 bit.
If software performs an mtc0 instruction which writes a non-zero value to an R/W0 field, the write to the R/W0 field will be ignored, but permitted writes to other fields in the register will not be affected.
Software can only clear an R/W0 bit.
Software writes 0 to an R/W0 field to clear the field.
Software writes nonzero to an R/W0 bit in order to guarantee that the bit is not affected by the write.
1.5 For More Information
MIPS processor manuals and additional information about MIPS products can be found at http://www.mips.com.
. .
9
The MIPS32® Instruction Set Manual, Revision 6.06

Chapter 2
Guide to the Instruction Set
This chapter provides a detailed guide to understanding the instruction descriptions, which are listed in alphabetical order in the tables at the beginning of the next chapter.
2.1 Understanding the Instruction Fields
Figure 2.1 shows an example instruction. Following the figure are descriptions of the fields listed below:
• “Instruction Fields” on page 12
• “Instruction Descriptive Name and Mnemonic” on page 12
• “Format Field” on page 12
• “Purpose Field” on page 13
• “Description Field” on page 13
• “Restrictions Field” on page 13
• “Operation Field” on page 15
• “Exceptions Field” on page 15
• “Programming Notes and Implementation Notes Fields” on page 15
The MIPS32® Instruction Set Manual, Revision 6.06 10

2.1.1 Instruction Fields
Fields encoding the instruction word are shown in register form at the top of the instruction description. The follow- ing rules are followed:
• The values of constant fields and the opcode names are listed in uppercase (SPECIAL and ADD in Figure 2.2). Constant values in a field are shown in binary below the symbolic or hexadecimal value.
• All variable fields are listed with the lowercase names used in the instruction description (rs, rt, and rd in Figure 2.2).
• Fields that contain zeros but are not named are unused fields that are required to be zero (bits 10:6 in Figure 2.2). If such fields are set to non-zero values, the operation of the processor is UNPREDICTABLE.
Figure 2.2 Example of Instruction Fields
2.1.2 Instruction Descriptive Name and Mnemonic
The instruction descriptive name and mnemonic are printed as page headings for each instruction, as shown in Figure 2.3.
Figure 2.3 Example of Instruction Descriptive Name and Mnemonic
2.1.3 Format Field
The assembler formats for the instruction and the architecture level at which the instruction was originally defined are given in the Format field. If the instruction definition was later extended, the architecture levels at which it was extended and the assembler formats for the extended definition are shown in their order of extension (for an example, see C.cond fmt). The MIPS architecture levels are inclusive; higher architecture levels include all instructions in pre- vious levels. Extensions to instructions are backwards compatible. The original assembler formats are valid for the extended architecture.
Figure 2.4 Example of Instruction Format
The assembler format is shown with literal parts of the assembler instruction printed in uppercase characters. The variable parts, the operands, are shown as the lowercase names of the appropriate fields.
The architectural level at which the instruction was first defined, for example “MIPS32” is shown at the right side of the page. Instructions introduced at different times by different ISA family members, are indicated by markings such
2.1 Understanding the Instruction Fields
31 26 25 21 20 16 15 11 10 6 5 0
SPECIAL 000000
rs
rt
rd
0 00000
ADD 100000
655556
Add Word ADD
Format: ADD fd,rs,rt MIPS32
The MIPS32® Instruction Set Manual, Revision 6.06 12

Guide to the Instruction Set
as “MIPS64, MIPS32 Release 2”. Instructions removed by particular architecture release are indicated in the Avail- ability section.
There can be more than one assembler format for each architecture level. Floating point operations on formatted data show an assembly format with the actual assembler mnemonic for each valid value of the fmt field. For example, the ADD fmt instruction lists both ADD.S and ADD.D.
The assembler format lines sometimes include parenthetical comments to help explain variations in the formats (once again, see C.cond.fmt). These comments are not a part of the assembler format.
2.1.4 Purpose Field
The Purpose field gives a short description of the use of the instruction. Figure 2.5 Example of Instruction Purpose
2.1.5 Description Field
If a one-line symbolic description of the instruction is feasible, it appears immediately to the right of the Description heading. The main purpose is to show how fields in the instruction are used in the arithmetic or logical operation.
Figure 2.6 Example of Instruction Description
Purpose: Add Word
To add 32-bit integers. If an overflow occurs, then trap.
Description: GPR[rd]  GPR[rs] + GPR[rt]
The 32-bit word value in GPR rt is added to the 32-bit value in GPR rs to produce a 32-bit
result.
• If the addition results in 32-bit 2’s complement arithmetic overflow, the destination register is not modified and an Integer Overflow exception occurs.
• If the addition does not overflow, the 32-bit result is placed into GPR rd.
13
The MIPS32® Instruction Set Manual, Revision 6.06
The body of the section is a description of the operation of the instruction in text, tables, and figures. This description complements the high-level language description in the Operation section.
This section uses acronyms for register descriptions. “GPR rt” is CPU general-purpose register specified by the instruction field rt. “FPR fs” is the floating point operand register specified by the instruction field fs. “CP1 register fd” is the coprocessor 1 general register specified by the instruction field fd. “FCSR” is the floating point Control / Status register.
2.1.6 Restrictions Field
The Restrictions field documents any possible restrictions that may affect the instruction. Most restrictions fall into one of the following six categories:
• Valid values for instruction fields (for example, see floating point ADD.fmt)

• ALIGNMENT requirements for memory addresses (for example, see LW)
• Valid values of operands (for example, see ALNV.PS)
• Valid operand formats (for example, see floating point ADD.fmt)
• Order of instructions necessary to guarantee correct execution. These ordering constraints avoid pipeline hazards for which some processors do not have hardware interlocks (for example, see MUL).
• Valid memory access types (for example, see LL/SC)
Figure 2.7 Example of Instruction Restrictions
2.1.7 Availability and Compatibility Fields
The Availability and Compatibility sections are not provided for all instructions. These sections list considerations relevant to whether and how an implementation may implement some instructions, when software may use such instructions, and how software can determine if an instruction or feature is present. Such considerations include:
• Some instructions are not present on all architecture releases. Sometimes the implementation is required to signal a Reserved Instruction exception, but sometimes executing such an instruction encoding is architec- turally defined to give UNPREDICTABLE results.
• Some instructions are available for implementations of a particular architecture release, but may be provided only if an optional feature is implemented. Control register bits typically allow software to determine if the feature is present.
• Some instructions may not behave the same way on all implementations. Typically this involves behavior that was UNPREDICTABLE in some implementations, but which is made architectural and guaranteed con- sistent so that software can rely on it in subsequent architecture releases.
• Some instructions are prohibited for certain architecture releases and/or optional feature combinations.
• Some instructions may be removed for certain architecture releases. Implementations may then be required to signal a Reserved Instruction exception for the removed instruction encoding; but sometimes the instruc- tion encoding is reused for other instructions.
All of these considerations may apply to the same instruction. If such considerations applicable to an instruction are simple, the architecture level in which an instruction was defined or redefined in the Format field, and/or the Restric- tions section, may be sufficient; but if the set of such considerations applicable to an instruction is complicated, the Availability and Compatibility sections may be provided.
2.1 Understanding the Instruction Fields
Restrictions:
None
The MIPS32® Instruction Set Manual, Revision 6.06 14

Guide to the Instruction Set 2.1.8 Operation Field
The Operation field describes the operation of the instruction as pseudocode in a high-level language notation resem- bling Pascal. This formal description complements the Description section; it is not complete in itself because many of the restrictions are either difficult to include in the pseudocode or are omitted for legibility.
Figure 2.8 Example of Instruction Operation
Operation:
temp  (GPR[rs]31||GPR[rs]31..0) + (GPR[rt]31||GPR[rt]31..0) if temp32  temp31 then
SignalException(IntegerOverflow)
else
GPR[rd]  temp endif
15
The MIPS32® Instruction Set Manual, Revision 6.06
See 2.2 “Operation Section Notation and Functions” on page 16 for more information on the formal notation used here.
2.1.9 Exceptions Field
The Exceptions field lists the exceptions that can be caused by Operation of the instruction. It omits exceptions that can be caused by the instruction fetch, for instance, TLB Refill, and also omits exceptions that can be caused by asyn- chronous external events such as an Interrupt. Although a Bus Error exception may be caused by the operation of a load or store instruction, this section does not list Bus Error for load and store instructions because the relationship between load and store instructions and external error indications, like Bus Error, are dependent upon the implemen- tation.
Figure 2.9 Example of Instruction Exception
An instruction may cause implementation-dependent exceptions that are not present in the Exceptions section. 2.1.10 Programming Notes and Implementation Notes Fields
The Notes sections contain material that is useful for programmers and implementors, respectively, but that is not necessary to describe the instruction and does not belong in the description sections.
Exceptions:
Integer Overflow

2.2 Operation Section Notation and Functions Figure 2.10 Example of Instruction Programming Notes
2.2 Operation Section Notation and Functions
In an instruction description, the Operation section uses a high-level language notation to describe the operation per- formed by each instruction. Special symbols used in the pseudocode are described in the previous chapter. Specific pseudocode functions are described below.
This section presents information about the following topics:
• “Instruction Execution Ordering” on page 16
• “Pseudocode Functions” on page 16
2.2.1 Instruction Execution Ordering
Each of the high-level language statements in the Operations section are executed sequentially (except as constrained by conditional and loop constructs).
2.2.2 Pseudocode Functions
There are several functions used in the pseudocode descriptions. These are used either to make the pseudocode more readable, to abstract implementation-specific behavior, or both. These functions are defined in this section, and include the following:
• “Coprocessor General Register Access Functions” on page 16
• “Memory Operation Functions” on page 18
• “Floating Point Functions” on page 21
• “Miscellaneous Functions” on page 25
2.2.2.1 Coprocessor General Register Access Functions
Defined coprocessors, except for CP0, have instructions to exchange words and doublewords between coprocessor general registers and the rest of the system. What a coprocessor does with a word or doubleword supplied to it and how a coprocessor supplies a word or doubleword is defined by the coprocessor itself. This behavior is abstracted into the functions described in this section.
2.2.2.1.1 COP_LW
The COP_LW function defines the action taken by coprocessor z when supplied with a word from memory during a load word operation. The action is coprocessor-specific. The typical action would be to store the contents of mem- word in coprocessor general register rt.
Figure 2.11 COP_LW Pseudocode Function
COP_LW (z, rt, memword)
Programming Notes:
ADDU performs the same arithmetic operation but does not trap on overflow.
The MIPS32® Instruction Set Manual, Revision 6.06 16

Guide to the Instruction Set
17
The MIPS32® Instruction Set Manual, Revision 6.06
z: The coprocessor unit number
rt: Coprocessor general register specifier
memword: A 32-bit word value supplied to the coprocessor
/* Coprocessor-dependent action */
endfunction COP_LW
2.2.2.1.2 COP_LD
The COP_LD function defines the action taken by coprocessor z when supplied with a doubleword from memory during a load doubleword operation. The action is coprocessor-specific. The typical action would be to store the con- tents of memdouble in coprocessor general register rt.
Figure 2.12 COP_LD Pseudocode Function
COP_LD (z, rt, memdouble)
z: The coprocessor unit number
rt: Coprocessor general register specifier
memdouble: 64-bit doubleword value supplied to the coprocessor.
/* Coprocessor-dependent action */
endfunction COP_LD
2.2.2.1.3 COP_SW
The COP_SW function defines the action taken by coprocessor z to supply a word of data during a store word opera- tion. The action is coprocessor-specific. The typical action would be to supply the contents of the low-order word in coprocessor general register rt.
Figure 2.13 COP_SW Pseudocode Function
dataword  COP_SW (z, rt)
z: The coprocessor unit number
rt: Coprocessor general register specifier dataword: 32-bit word value
/* Coprocessor-dependent action */
endfunction COP_SW
2.2.2.1.4 COP_SD
The COP_SD function defines the action taken by coprocessor z to supply a doubleword of data during a store dou- bleword operation. The action is coprocessor-specific. The typical action would be to supply the contents of the low- order doubleword in coprocessor general register rt.
Figure 2.14 COP_SD Pseudocode Function
datadouble  COP_SD (z, rt)
z: The coprocessor unit number
rt: Coprocessor general register specifier datadouble: 64-bit doubleword value
/* Coprocessor-dependent action */

endfunction COP_SD
2.2.2.1.5 CoprocessorOperation
The CoprocessorOperation function performs the specified Coprocessor operation.
Figure 2.15 CoprocessorOperation Pseudocode Function
CoprocessorOperation (z, cop_fun)
/* z: Coprocessor unit number */
/* cop_fun: Coprocessor function from function field of instruction */
/* Transmit the cop_fun value to coprocessor z */ endfunction CoprocessorOperation
2.2.2.2 Memory Operation Functions
Regardless of byte ordering (big- or little-endian), the address of a halfword, word, or doubleword is the smallest byte address of the bytes that form the object. For big-endian ordering this is the most-significant byte; for a little-endian ordering this is the least-significant byte.
In the Operation pseudocode for load and store operations, the following functions summarize the handling of virtual addresses and the access of physical memory. The size of the data item to be loaded or stored is passed in the Access- Length field. The valid constant names and values are shown in Table 2.1. The bytes within the addressed unit of memory (word for 32-bit processors or doubleword for 64-bit processors) that are used can be determined directly from the AccessLength and the two or three low-order bits of the address.
2.2.2.2.1 Misaligned Support
MIPS processors originally required all memory accesses to be naturally aligned. MSA (the MIPS SIMD Architec- ture) supported misaligned memory accesses for its 128 bit packed SIMD vector loads and stores, from its introduc- tion in MIPS Release 5. Release 6 requires systems to provide support for misaligned memory accesses for all ordinary memory reference instructions: the system must provide a mechanism to complete a misaligned memory ref- erence for this instruction, ranging from full execution in hardware to trap-and-emulate.
The pseudocode function MisalignedSupport encapsulates the version number check to determine if misalignment is supported for an ordinary memory access.
Figure 2.16 MisalignedSupport Pseudocode Function
predicate  MisalignedSupport ()
return Config.AR ≥ 2 // Architecture Revision 2 corresponds to MIPS Release 6.
end function
See Appendix B, “Misaligned Memory Accesses” on page 511 for a more detailed discussion of misalignment, including pseudocode functions for the actual misaligned memory access.
2.2.2.2.2 AddressTranslation
The AddressTranslation function translates a virtual address to a physical address and its cacheability and coherency attribute, describing the mechanism used to resolve the memory reference.
The MIPS32® Instruction Set Manual, Revision 6.06 18
2.2 Operation Section Notation and Functions

Guide to the Instruction Set
19
The MIPS32® Instruction Set Manual, Revision 6.06
Given the virtual address vAddr, and whether the reference is to Instructions or Data (IorD), find the corresponding physical address (pAddr) and the cacheability and coherency attribute (CCA) used to resolve the reference. If the vir- tual address is in one of the unmapped address spaces, the physical address and CCA are determined directly by the virtual address. If the virtual address is in one of the mapped address spaces then the TLB or fixed mapping MMU determines the physical address and access type; if the required translation is not present in the TLB or the desired access is not permitted, the function fails and an exception is taken.
Figure 2.17 AddressTranslation Pseudocode Function
(pAddr, CCA)  AddressTranslation (vAddr, IorD, LorS)
/* pAddr: physical address */
/* CCA: Cacheability&Coherency Attribute,the method used to access caches*/ /* and memory and resolve the reference */
/* vAddr: virtual address */
/* IorD: Indicates whether access is for INSTRUCTION or DATA */ /* LorS: Indicates whether access is for LOAD or STORE */
/* See the address translation description for the appropriate MMU */
/* type in Volume III of this book for the exact translation mechanism */
endfunction AddressTranslation
2.2.2.2.3 LoadMemory
The LoadMemory function loads a value from memory.
This action uses cache and main memory as specified in both the Cacheability and Coherency Attribute (CCA) and the access (IorD) to find the contents of AccessLength memory bytes, starting at physical location pAddr. The data is returned in a fixed-width naturally aligned memory element (MemElem). The low-order 2 (or 3) bits of the address and the AccessLength indicate which of the bytes within MemElem need to be passed to the processor. If the memory access type of the reference is uncached, only the referenced bytes are read from memory and marked as valid within the memory element. If the access type is cached but the data is not present in cache, an implementation-specific size and alignment block of memory is read and loaded into the cache to satisfy a load reference. At a minimum, this block is the entire memory element.
Figure 2.18 LoadMemory Pseudocode Function
MemElem  LoadMemory (CCA, AccessLength, pAddr, vAddr, IorD)
/* MemElem: /*
/*
/*
/* CCA: /*
Data is returned in a fixed width with a natural alignment. The */
width is the same size as the CPU general-purpose register, */
32 or 64 bits, aligned on a 32- or 64-bit boundary, */
respectively. */
Cacheability&CoherencyAttribute=method used to access caches */
and memory and resolve the reference */
/* AccessLength: Length, in bytes, of access */
/* pAddr:
/* vAddr:
/* IorD:
physical address */
virtual address */
Indicates whether access is for Instructions or Data */
endfunction LoadMemory

2.2.2.2.4 StoreMemory
The StoreMemory function stores a value to memory.
The specified data is stored into the physical location pAddr using the memory hierarchy (data caches and main mem- ory) as specified by the Cacheability and Coherency Attribute (CCA). The MemElem contains the data for an aligned, fixed-width memory element (a word for 32-bit processors, a doubleword for 64-bit processors), though only the bytes that are actually stored to memory need be valid. The low-order two (or three) bits of pAddr and the AccessLen- gth field indicate which of the bytes within the MemElem data should be stored; only these bytes in memory will actually be changed.
Figure 2.19 StoreMemory Pseudocode Function
StoreMemory (CCA, AccessLength, MemElem, pAddr, vAddr)
/* CCA: Cacheability&Coherency Attribute, the method used to access */ /* caches and memory and resolve the reference. */
/* AccessLength: Length, in bytes, of access */
/* MemElem: /*
/*
/*
/* /*
/* pAddr:
/* vAddr:
Data in the width and alignment of a memory element. */
The width is the same size as the CPU general */
purpose register, either 4 or 8 bytes, */
aligned on a 4- or 8-byte boundary. For a */
partial-memory-element store, only the bytes that will be*/
stored must be valid.*/
physical address */
virtual address */
endfunction StoreMemory
2.2.2.2.5 Prefetch
The Prefetch function prefetches data from memory.
Prefetch is an advisory instruction for which an implementation-specific action is taken. The action taken may increase performance but must not change the meaning of the program or alter architecturally visible state.
Figure 2.20 Prefetch Pseudocode Function
Prefetch (CCA, pAddr, vAddr, DATA, hint)
/* CCA: Cacheability&Coherency Attribute, the method used to access */ /* caches and memory and resolve the reference. */
/* pAddr: physical address */
/* vAddr: virtual address */
/* DATA: Indicates that access is for DATA */
/* hint: hint that indicates the possible use of the data */
endfunction Prefetch
Table 2.1 lists the data access lengths and their labels for loads and stores.
Table 2.1 AccessLength Specifications for Loads/Stores
2.2 Operation Section Notation and Functions
AccessLength Name
Value
Meaning
DOUBLEWORD
7
8 bytes (64 bits)
The MIPS32® Instruction Set Manual, Revision 6.06 20

Guide to the Instruction Set
Table 2.1 AccessLength Specifications for Loads/Stores
AccessLength Name
Value
Meaning
SEPTIBYTE
6
7 bytes (56 bits)
SEXTIBYTE
5
6 bytes (48 bits)
QUINTIBYTE
4
5 bytes (40 bits)
WORD
3
4 bytes (32 bits)
TRIPLEBYTE
2
3 bytes (24 bits)
HALFWORD
1
2 bytes (16 bits)
BYTE
0
1 byte (8 bits)
21
The MIPS32® Instruction Set Manual, Revision 6.06
2.2.2.2.6 SyncOperation
The SyncOperation function orders loads and stores to synchronize shared memory.
This action makes the effects of the synchronizable loads and stores indicated by stype occur in the same order for all processors.
Figure 2.21 SyncOperation Pseudocode Function
SyncOperation(stype)
/* stype: Type of load/store ordering to perform. */
/* Perform implementation-dependent operation to complete the */
/* required synchronization operation */
endfunction SyncOperation
2.2.2.3 Floating Point Functions
The pseudocode shown in below specifies how the unformatted contents loaded or moved to CP1 registers are inter- preted to form a formatted value. If an FPR contains a value in some format, rather than unformatted contents from a load (uninterpreted), it is valid to interpret the value in that format (but not to interpret it in a different format).
2.2.2.3.1 ValueFPR
The ValueFPR function returns a formatted value from the floating point registers.
Figure 2.22 ValueFPR Pseudocode Function
value  ValueFPR(fpr, fmt)
/* value: The formattted value from the FPR */
/* fpr:
/* fmt:
/*
/*
The FPR number */
The format of the data, one of: */
S, D, W, L, PS, */
OB, QH, */
UNINTERPRETED_WORD, */
UNINTERPRETED_DOUBLEWORD */
/*
/*
/* The UNINTERPRETED values are used to indicate that the datatype */
/* is not known as, for example, in SWC1 and SDC1 */

case fmt of
S, W, UNINTERPRETED_WORD:
valueFPR  FPR[fpr]
D, UNINTERPRETED_DOUBLEWORD: if (FP32RegistersMode  0)
if (fpr0  0) then
valueFPR  UNPREDICTABLE
else
valueFPR  FPR[fpr1]31..0  FPR[fpr]31..0
endif else
valueFPR  FPR[fpr] endif
L:
if (FP32RegistersMode  0) then
valueFPR  UNPREDICTABLE else
valueFPR  FPR[fpr] endif
DEFAULT:
valueFPR  UNPREDICTABLE
endcase
endfunction ValueFPR
The pseudocode shown below specifies the way a binary encoding representing a formatted value is stored into CP1 registers by a computational or move operation. This binary representation is visible to store or move-from instruc- tions. Once an FPR receives a value from the StoreFPR(), it is not valid to interpret the value with ValueFPR() in a different format.
2.2.2.3.2 StoreFPR
Figure 2.23 StoreFPR Pseudocode Function
StoreFPR (fpr, fmt, value)
/* fpr:
/* fmt:
/*
/*
The FPR number */
The format of the data, one of: */
S, D, W, L, PS, */
OB, QH, */
UNINTERPRETED_WORD, */
UNINTERPRETED_DOUBLEWORD */
/*
/*
/* value: The formattted value to be stored into the FPR */
/* The UNINTERPRETED values are used to indicate that the datatype */
/* is not known as, for example, in LWC1 and LDC1 */
case fmt of
S, W, UNINTERPRETED_WORD:
FPR[fpr]  value
D, UNINTERPRETED_DOUBLEWORD:
The MIPS32® Instruction Set Manual, Revision 6.06 22
2.2 Operation Section Notation and Functions

Guide to the Instruction Set
23
The MIPS32® Instruction Set Manual, Revision 6.06
if (FP32RegistersMode  0) if (fpr0  0) then
UNPREDICTABLE
else
FPR[fpr]  UNPREDICTABLE32  value FPR[fpr1]  UNPREDICTABLE32  value
endif else
FPR[fpr]  value endif
L:
if (FP32RegistersMode  0) then
UNPREDICTABLE
else
FPR[fpr]  value
endif
endcase
endfunction StoreFPR
2.2.2.3.3 CheckFPException
The pseudocode shown below checks for an enabled floating point exception and conditionally signals the exception.
Figure 2.24 CheckFPException Pseudocode Function
CheckFPException()
/* A floating point exception is signaled if the E bit of the Cause field is a 1 */ /* (Unimplemented Operations have no enable) or if any bit in the Cause field */ /* and the corresponding bit in the Enable field are both 1 */
if ( (FCSR17  1) or
((FCSR16..12 and FCSR11..7)  0)) ) then
SignalException(FloatingPointException)
endif
endfunction CheckFPException
2.2.2.3.4 FPConditionCode
The FPConditionCode function returns the value of a specific floating point condition code.
Figure 2.25 FPConditionCode Pseudocode Function
tf FPConditionCode(cc)
/* tf: The value of the specified condition code */ /* cc: The Condition code number in the range 0..7 */
if cc = 0 then FPConditionCode  FCSR23
else
FPConditionCode  FCSR24+cc
31..0 63..32

endif
endfunction FPConditionCode
2.2.2.3.5 SetFPConditionCode
The SetFPConditionCode function writes a new value to a specific floating point condition code.
Figure 2.26 SetFPConditionCode Pseudocode Function
SetFPConditionCode(cc, tf)
if cc = 0 then
FCSR  FCSR31..24 || tf || FCSR22..0 else
FCSR  FCSR31..25+cc || tf || FCSR23+cc..0 endfunction SetFPConditionCode
2.2.2.4 Pseudocode Functions Related to Sign and Zero Extension 2.2.2.4.1 Sign extension and zero extension in pseudocode
Much pseudocode uses a generic function sign_extend without specifying from what bit position the extension is done, when the intention is obvious. E.g. sign_extend(immediate16) or sign_extend(disp9).
However, sometimes it is necessary to specify the bit position. For example, sign_extend(temp31..0) or the more complicated (offset15)GPRLEN-(16+2) || offset || 02.
The explicit notation sign_extend.nbits(val) or sign_extend(val,nbits) is suggested as a simpli- fication. They say to sign extend as if an nbits-sized signed integer. The width to be sign extended to is usually appar- ent by context, and is usually GPRLEN, 32 or 64 bits. The previous examples then become.
sign_extend(temp31..0)
= sign_extend.32(temp)
and
Note that sign_extend.N(value) extends from bit position N-1, if the bits are numbered 0..N-1 as is typical.
The explicit notations sign_extend.nbits(val) or sign_extend(val,nbits) is used as a simplifica- tion. These notations say to sign extend as if an nbits-sized signed integer. The width to be sign extended to is usually apparent by context, and is usually GPRLEN, 32 or 64 bits.
Figure 2.27 sign_extend Pseudocode Functions
sign_extend.nbits(val) = sign_extend(val,nbits) /* syntactic equivalents */
function sign_extend(val,nbits)
return (valnbits-1)GPRLEN-nbits || valnbits-1..0
end function
The earlier examples can be expressed as
endif
(offset15)GPRLEN-(16+2) || offset || 02 = sign_extend.16(offset)<<2 The MIPS32® Instruction Set Manual, Revision 6.06 24 2.2 Operation Section Notation and Functions Guide to the Instruction Set 25 The MIPS32® Instruction Set Manual, Revision 6.06 (offset15)GPRLEN-(16+2) || offset || 02 = sign_extend.16(offset) << 2) and Similarly for zero_extension, although zero extension is less common than sign extension in the MIPS ISA. Floating point may use notations such as zero_extend.fmt corresponding to the format of the FPU instruction. E.g. zero_extend.S and zero_extend.D are equivalent to zero_extend.32 and zero_extend.64. Existing pseudocode may use any of these, or other, notations. 2.2.2.4.2 memory_address The pseudocode function memory_address performs mode-dependent address space wrapping for compatibility between MIPS32 and MIPS64. It is applied to all memory references. It may be specified explicitly in some places, particularly for new memory reference instructions, but it is also declared to apply implicitly to all memory refer- ences as defined below. In addition, certain instructions that are used to calculate effective memory addresses but which are not themselves memory accesses specify memory_address explicitly in their pseudocode. Figure 2.28 memory_address Pseudocode Function function memory_address(ea) return ea end function On a 32-bit CPU, memory_address returns its 32-bit effective address argument unaffected. In addition to the use of memory_address for all memory references (including load and store instructions, LL/ SC), Release 6 extends this behavior to control transfers (branch and call instructions), and to the PC-relative address calculation instructions (ADDIUPC, AUIPC, ALUIPC). In newer instructions the function is explicit in the pseudo- code. Implicit address space wrapping for all instruction fetches is described by the following pseudocode fragment which should be considered part of instruction fetch: Figure 2.29 Instruction Fetch Implicit memory_address Wrapping PC  memory_address( PC ) ( instruction_data, length )  instruction_fetch( PC ) /* decode and execute instruction */ Implicit address space wrapping for all data memory accesses is described by the following pseudocode, which is inserted at the top of the AddressTranslation pseudocode function: Figure 2.30 AddressTranslation implicit memory_address Wrapping (pAddr, CCA)  AddressTranslation (vAddr, IorD, LorS) vAddr  memory_address(vAddr) In addition to its use in instruction pseudocode, 2.2.2.5 Miscellaneous Functions This section lists miscellaneous functions not covered in previous sections. sign_extend(temp31..0) = sign_extend.32(temp) 2.2.2.5.1 SignalException The SignalException function signals an exception condition. This action results in an exception that aborts the instruction. The instruction operation pseudocode never sees a return from this function call. Figure 2.31 SignalException Pseudocode Function SignalException(Exception, argument) /* Exception: The exception condition that exists. */ /* argument: A exception-dependent argument, if any */ endfunction SignalException 2.2.2.5.2 SignalDebugBreakpointException The SignalDebugBreakpointException function signals a condition that causes entry into Debug Mode from non- Debug Mode. This action results in an exception that aborts the instruction. The instruction operation pseudocode never sees a return from this function call. Figure 2.32 SignalDebugBreakpointException Pseudocode Function SignalDebugBreakpointException() endfunction SignalDebugBreakpointException 2.2.2.5.3 SignalDebugModeBreakpointException The SignalDebugModeBreakpointException function signals a condition that causes entry into Debug Mode from Debug Mode (i.e., an exception generated while already running in Debug Mode). This action results in an exception that aborts the instruction. The instruction operation pseudocode never sees a return from this function call. Figure 2.33 SignalDebugModeBreakpointException Pseudocode Function SignalDebugModeBreakpointException() endfunction SignalDebugModeBreakpointException 2.2.2.5.4 NullifyCurrentInstruction The NullifyCurrentInstruction function nullifies the current instruction. The instruction is aborted, inhibiting not only the functional effect of the instruction, but also inhibiting all exceptions detected during fetch, decode, or execution of the instruction in question. For branch-likely instructions, nullification kills the instruction in the delay slot of the branch likely instruction. Figure 2.34 NullifyCurrentInstruction PseudoCode Function NullifyCurrentInstruction() The MIPS32® Instruction Set Manual, Revision 6.06 26 2.2 Operation Section Notation and Functions Guide to the Instruction Set endfunction NullifyCurrentInstruction 2.2.2.5.5 PolyMult The PolyMult function multiplies two binary polynomial coefficients. Figure 2.35 PolyMult Pseudocode Function PolyMult(x, y) temp  0 for i in 0 .. 31 if xi = 1 then temp  temp xor (y (31-i)..0 endif endfor PolyMult  temp endfunction PolyMult || 0i) 2.3 Op and Function Subfield Notation In some instructions, the instruction subfields op and function can have constant 5- or 6-bit values. When reference is made to these instructions, uppercase mnemonics are used. For instance, in the floating point ADD instruction, op=COP1 and function=ADD. In other cases, a single field has both fixed and variable subfields, so the name con- tains both upper- and lowercase characters. 2.4 FPU Instructions In the detailed description of each FPU instruction, all variable subfields in an instruction format (such as fs, ft, imme- diate, and so on) are shown in lowercase. The instruction name (such as ADD, SUB, and so on) is shown in upper- case. For the sake of clarity, an alias is sometimes used for a variable subfield in the formats of specific instructions. For example, rs=base in the format for load and store instructions. Such an alias is always lowercase since it refers to a variable subfield. Bit encodings for mnemonics are given in Volume I, in the chapters describing the CPU, FPU, MDMX, and MIPS16e instructions. See “Op and Function Subfield Notation” on page 27 for a description of the op and function subfields. 27 The MIPS32® Instruction Set Manual, Revision 6.06 The MIPS32® Instruction Set Manual, Revision 6.06 28 2.4 FPU Instructions Chapter 3 The MIPS32® Instruction Set 3.1 Compliance and Subsetting To be compliant with the MIPS32 Architecture, designs must implement a set of required features, as described in this document set. To allow implementation flexibility, the MIPS32 Architecture provides subsetting rules. An imple- mentation that follows these rules is compliant with the MIPS32 Architecture as long as it adheres strictly to the rules, and fully implements the remaining instructions. Supersetting of the MIPS32 Architecture is only allowed by adding functions to the SPECIAL2, COP2, or both major opcodes, by adding control for co-processors via the COP2, LWC2, SWC2, LDC2, and/or SDC2, or via the addition of approved Application Specific Extensions. Release 6 removes all instructions under the SPECIAL2 major opcode, either by removing them or moving them to the COP2 major opcode. All coprocessor 2 support instructions (for example, LWC2) have been moved to the COP2 major opcode. Supersetting of the Release 6 architecture is only allowed in the COP2 major opcode, or via the addi- tion of approved Application Specific Extensions. SPECIAL2 is reserved for MIPS. Note: The use of COP3 as a customizable coprocessor has been removed in the Release 2 of the MIPS32 architecture. The COP3 is reserved for the future extension of the architecture. Implementations using Release1 of the MIPS32 architecture are strongly discouraged from using the COP3 opcode for a user-available coprocessor as doing so will limit the potential for an upgrade path to a 64-bit floating point unit. The instruction set subsetting rules are described in the subsections below, and also the following rule: • Co-dependence of Architecture Features: MIPSr5TM (also called Release 5) and subsequent releases (such as Release 6) include a number of features. Some are optional; some are required. Features provided by a release, such as MIPSr5 or later, whether optional or required, must be consistent. If any feature that is introduced by a particular release is implemented (such as a feature described as part of Release 5 and not any earlier release) then all other features must be implemented in a manner consistent with that release. For example: the VZ and MSA features are introduced by Release 5 but are optional. The FR=1 64-bit FPU register model was optional when introduced earlier, but is now required by Release 5 if any FPU is implemented. If any or all of VZ or MSA are implemented, then Release 5 is implied, and then if an FPU is implemented, it must implement the FR=1 64- bit FPU register model. 3.1.1 Subsetting of Non-Privileged Architecture • All non-privileged (do not need access to Coprocessor 0) CPU (non-FPU) instructions must be implemented — no subsetting of these are allowed — per the MIPS Instruction Set Architecture release supported. • If any instruction is subsetted out based on the rules below, an attempt to execute that instruction must cause the appropriate exception (typically Reserved Instruction or Coprocessor Unusable). • The FPU and related support instructions, such as CPU conditional branches on FPU conditions (pre-Release 6 BC1T/BC1F, Release 6 BC1NEQZ) and CPU conditional moves on FPU conditions (pre-Release 6 MOVT/ MOVF), may be omitted. Software may determine if an FPU is implemented by checking the state of the FP bit in the Config1 CP0 register. Software may determine which FPU data types are implemented by checking the The MIPS32® Instruction Set Manual, Revision 6.06 29 appropriate bits in the FIR CP1 register. The following allowable FPU subsets are compliant with the MIPS32 architecture: • No FPU Config1.FP=0 • FPU with S, and W formats and all supporting instructions. This 32-bit subset is permitted by Release 6, but prohibited by pre-Release 6 releases. Config1.FP=1, Status.FR=0, FIR.S=FIR.L=1, FIR.D=FIR.L=FIR.PS=0. • FPU with S, D, W, and L formats and all supporting instructions Config1.FP=1, Status.FR=(see below), FIR.S=FIR.L=FIR.D=FIR.L=1, FIR.PS=0. pre-MIPSr5 permits this 64-bit configuration, and allows both FPU register modes. Status.FR=0 support is required but Status.FR=1 support is optional. MIPSr5 permits this 64-bit configuration, and requires both FPU register modes, i.e. both Status.FR=0 and Status.FR=1 support are required. Release 6 permits this 64-bit configuration, but requires Status.FR=1 and FIR.F64=1. Release 6 prohibits Status.FR=0 if FIR.D=1 or FIR.L=1. • FPU with S, D, PS, W, and L formats and all supporting instructions Config1.FP=1, Status.FR=0/1, FIR.S=FIR.L=FIR.D=FIR.L=FIR.PS=1. Release 6 prohibits this mode, and any mode with FIR.PS=1 paired single support. • In Release 5 of the Architecture, if floating point is implemented then FR=1 is required. I.e. the 64-bit FPU, with the FR=1 64-bit FPU register model, is required. The FR=0 32-bit FPU register model continues to be required. • Coprocessor 2 is optional and may be omitted. Software may determine if Coprocessor 2 is implemented by checking the state of the C2 bit in the Config1 CP0 register. If Coprocessor 2 is implemented, the Coprocessor 2 interface instructions (BC2, CFC2, COP2, CTC2, LDC2, LWC2, MFC2, MTC2, SDC2, and SWC2) may be omitted on an instruction-by-instruction basis. • The caches are optional. The Config1DL and Config1IL fields denote whether the first level caches are present or not. • Instruction, CP0 Register, and CP1 Control Register fields that are marked “Reserved” or shown as “0” in the description of that field are reserved for future use by the architecture and are not available to implementations. Implementations may only use those fields that are explicitly reserved for implementation dependent use. • Supported Modules/ASEs are optional and may be subsetted out. In most cases, software may determine if a sup- ported Module/ASE is implemented by checking the appropriate bit in the Config1 or Config3 or Config4 CP0 register. If they are implemented, they must implement the entire ISA applicable to the component, or implement subsets that are approved by the Module/ASE specifications. The MIPS32® Instruction Set Manual, Revision 6.06 30 3.1 Compliance and Subsetting The MIPS32® Instruction Set • EJTAG is optional and may be subsetted out. If it is implemented, it must implement only those subsets that are approved by the EJTAG specification. If EJTAG is not implemented, the EJTAG instructions (SDBBP and DERET) can be subsetted out. • In MIPS Release 3, there are two architecture branches (MIPS32/64 and microMIPS32/64). A single device is allowed to implement both architecture branches. The Privileged Resource Architecture (COP0) registers do not mode-switch in width (32-bit vs. 64-bit). For this reason, if a device implements both architecture branches, the address/data widths must be consistent. If a device implements MIPS64 and also implements microMIPS, it must implement microMIPS64 not just microMIPS32. Simiarly, If a device implements microMIPS64 and also imple- ments MIPS32/64, it must implement MIPS64 not just MIPS32. • Prior to Release 6, the JALX instruction is required if and only if ISA mode-switching is possible. If both of the architecture branches are implemented (MIPS32/64 and microMIPS32/64) or if MIPS16e is implemented then the JALX instructions are required. If only one branch of the architecture family and MIPS16e is not imple- mented then the JALX instruction is not implemented. The JALX instruction was removed in Release 6. 3.2 Alphabetical List of Instructions The following pages present detailed descriptions of instructions, arranged alphabetical order of opcode mnemonic (except where several similar instructions are described together.) 31 The MIPS32® Instruction Set Manual, Revision 6.06 ABS.fmt IFloating Point Absolute Value 31 26 25 21 20 16 15 11 10 6 5 655556 COP1 010001 fmt 0 00000 fs fd ABS 000101 Format: ABS.fmt ABS.S fd, fs ABS.D fd, fs ABS.PS fd, fs Purpose: Floating Point Absolute Value Description: FPR[fd]  abs(FPR[fs]) 0 MIPS32 MIPS32 MIPS64,MIPS32 Release 2, removed in Release 6 The absolute value of the value in FPR fs is placed in FPR fd. The operand and result are values in format fmt. ABS.PS takes the absolute value of the two values in FPR fs independently, and ORs together any generated excep- tions. The Cause bits are ORed into the Flag bits if no exception is taken. If FIRHas2008=0 or FCSRABS2008=0 then this operation is arithmetic. For this case, any NaN operand signals invalid operation. If FCSRABS2008=1 then this operation is non-arithmetic. For this case, both regular floating point numbers and NAN values are treated alike, only the sign bit is affected by this instruction. No IEEE exception can be generated for this case, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPRE- DICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. The result of ABS.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. ABS.PS is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: ABS.PS has been removed in Release 6. Operation: StoreFPR(fd, fmt, AbsoluteValue(ValueFPR(fs, fmt))) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Unimplemented Operation, Invalid Operation The MIPS32® Instruction Set Manual, Revision 6.06 32 ADD Add Word 31 26 25 21 20 16 15 11 10 6 5 655556 Format: ADD rd, rs, rt Purpose: Add Word To add 32-bit integers. If an overflow occurs, then trap. Description: GPR[rd]  GPR[rs] + GPR[rt] The 32-bit word value in GPR rt is added to the 32-bit value in GPR rs to produce a 32-bit result. 0 MIPS32 SPECIAL 000000 rs rt rd 0 00000 ADD 100000 • If the addition results in 32-bit 2’s complement arithmetic overflow, the destination register is not modified and an Integer Overflow exception occurs. • If the addition does not overflow, the 32-bit result is placed into GPR rd. Restrictions: None Operation: temp  (GPR[rs]31||GPR[rs]31..0) + (GPR[rt]31||GPR[rt]31..0) if temp32  temp31 then SignalException(IntegerOverflow) else GPR[rd]  temp endif Exceptions: Integer Overflow Programming Notes: ADDU performs the same arithmetic operation but does not trap on overflow. The MIPS32® Instruction Set Manual, Revision 6.06 33 ADD.fmt Floating Point Add 31 26 25 21 20 16 15 11 10 6 5 655556 COP1 010001 fmt ft fs fd ADD 000000 34 The MIPS32® Instruction Set Manual, Revision 6.06 Format: ADD.fmt ADD.S fd, fs, ft ADD.D fd, fs, ft ADD.PS fd, fs, ft Purpose: Floating Point Add To add floating point values. 0 MIPS32 MIPS32 MIPS64,MIPS32 Release 2, removed in Release 6 Description: FPR[fd]  FPR[fs] + FPR[ft] The value in FPR ft is added to the value in FPR fs. The result is calculated to infinite precision, rounded by using to the current rounding mode in FCSR, and placed into FPR fd. The operands and result are values in format fmt. ADD.PS adds the upper and lower halves of FPR fs and FPR ft independently, and ORs together any generated excep- tions. The Cause bits are ORed into the Flag bits if no exception is taken. Restrictions: The fields fs, ft, and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPREDICTABLE. The operands must be values in format fmt. If the fields are not, the result is UNPREDICTABLE and the value of the operand FPRs becomes UNPREDICTABLE. The result of ADD.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. ADD.PS is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: ADD.PS has been removed in Release 6. Operation: StoreFPR (fd, fmt, ValueFPR(fs, fmt) fmt ValueFPR(ft, fmt)) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Unimplemented Operation, Invalid Operation, Inexact, Overflow, Underflow ADDI Add Immediate Word 31 26 25 21 20 16 15 655 16 Format: ADDI rt, rs, immediate Purpose: Add Immediate Word To add a constant to a 32-bit integer. If overflow occurs, then trap. Description: GPR[rt]  GPR[rs] + immediate 0 MIPS32, removed in Release 6 ADDI 001000 rs rt immediate The 16-bit signed immediate is added to the 32-bit value in GPR rs to produce a 32-bit result. • If the addition results in 32-bit 2’s complement arithmetic overflow, the destination register is not modified and an Integer Overflow exception occurs. • If the addition does not overflow, the 32-bit result is placed into GPR rt. Restrictions: Availability and Compatibility: This instruction has been removed in Release 6. The encoding has been reused for other instructions introduced by Release 6. Operation: temp  (GPR[rs]31||GPR[rs]31..0) + sign_extend(immediate) if temp32  temp31 then SignalException(IntegerOverflow) else GPR[rt]  temp endif Exceptions: Integer Overflow Programming Notes: ADDIU performs the same arithmetic operation but does not trap on overflow. The MIPS32® Instruction Set Manual, Revision 6.06 35 ADDIU Add Immediate Unsigned Word 31 26 25 21 20 16 15 0 ADDIU 001001 rs rt immediate 655 16 Format: ADDIU rt, rs, immediate Purpose: Add Immediate Unsigned Word To add a constant to a 32-bit integer. MIPS32 Description: GPR[rt]  GPR[rs] + immediate The 16-bit signed immediate is added to the 32-bit value in GPR rs and the 32-bit arithmetic result is placed into GPR rt. No Integer Overflow exception occurs under any circumstances. Restrictions: None Operation: temp  GPR[rs] + sign_extend(immediate) GPR[rt]  temp Exceptions: None Programming Notes: The term “unsigned” in the instruction name is a misnomer; this operation is 32-bit modulo arithmetic that does not trap on overflow. This instruction is appropriate for unsigned arithmetic, such as address arithmetic, or integer arith- metic environments that ignore overflow, such as C language arithmetic. The MIPS32® Instruction Set Manual, Revision 6.06 36 ADDIUPC Add Immediate to PC (unsigned - non-trapping) 31 2625 21201918 0 652 19 Format: ADDIUPC rs,immediate MIPS32 Release 6 Purpose: Add Immediate to PC (unsigned - non-trapping) Description: GPR[rs]  ( PC + sign_extend( immediate << 2 ) ) This instruction performs a PC-relative address calculation. The 19-bit immediate is shifted left by 2 bits, sign- extended, and added to the address of the ADDIUPC instruction. The result is placed in GPR rs. Restrictions: None Availability and Compatibility: This instruction is introduced by and required as of Release 6. Operation: GPR[rs]  ( PC + sign_extend( immediate << 2 ) ) Exceptions: None Programming Notes: The term “unsigned” in this instruction mnemonic is a misnomer. “Unsigned” here means “non-trapping”. It does not trap on a signed 32-bit overflow. ADDIUPC corresponds to unsigned ADDIU, which does not trap on overflow, as opposed to ADDI, which does trap on overflow. PCREL 111011 rs ADDIUPC 00 immediate The MIPS32® Instruction Set Manual, Revision 6.06 37 ADDU Add Unsigned Word 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: ADDU rd, rs, rt MIPS32 Purpose: Add Unsigned Word To add 32-bit integers. Description: GPR[rd]  GPR[rs] + GPR[rt] The 32-bit word value in GPR rt is added to the 32-bit value in GPR rs and the 32-bit arithmetic result is placed into GPR rd. No Integer Overflow exception occurs under any circumstances. Restrictions: None Operation: temp  GPR[rs] + GPR[rt] GPR[rd]  temp Exceptions: None Programming Notes: The term “unsigned” in the instruction name is a misnomer; this operation is 32-bit modulo arithmetic that does not trap on overflow. This instruction is appropriate for unsigned arithmetic, such as address arithmetic, or integer arith- metic environments that ignore overflow, such as C language arithmetic. SPECIAL 000000 rs rt rd 0 00000 ADDU 100001 38 The MIPS32® Instruction Set Manual, Revision 6.06 40 The MIPS32® Instruction Set Manual, Revision 6.06 ALIGN Concatenate two GPRs, and extract a contiguous subset at a byte position Exceptions: None ALNV.PS IFloating Point Align Variable StoreFPR(fd, PS,ValueFPR(fs,PS)) else if GPR[rs]2..0  4 then UNPREDICTABLE else if BigEndianCPU then StoreFPR(fd, PS, ValueFPR(fs, PS)31..0 || ValueFPR(ft,PS)63..32) else StoreFPR(fd, PS, ValueFPR(ft, PS)31..0 || ValueFPR(fs,PS)63..32) endif Exceptions: Coprocessor Unusable, Reserved Instruction Programming Notes: ALNV.PS is designed to be used with LUXC1 to load 8 bytes of data from any 4-byte boundary. For example: /* Copy T2 bytes (a multiple of 16) of data T0 to T1, T0 unaligned, T1 aligned. Reads one dw beyond the end of T0. */ LUXC1 F0, 0(T0) /* set up by reading 1st src dw */ LI T3, 0 /* index into src and dst arrays */ ADDIU T4, T0, 8 /* base for odd dw loads */ ADDIU T5, T1, -8/* base for odd dw stores */ LOOP: LUXC1 F1, T3(T4) ALNV.PS F2, F0, F1, T0/* switch F0, F1 for little-endian */ SDC1 F2, T3(T1) ADDIU T3, T3, 16 LUXC1 F0, T3(T0) ALNV.PS F2, F1, F0, T0/* switch F1, F0 for little-endian */ BNE T3, T2, LOOP SDC1 F2,T3(T5) DONE: ALNV.PS is also useful with SUXC1 to store paired-single results in a vector loop to a possibly misaligned address: /* T1[i] = T0[i] + F8, T0 aligned, T1 unaligned. */ CVT.PS.S F8, F8, F8/* make addend paired-single */ /* Loop header computes 1st pair into F0, stores high half if T1 */ /* misaligned */ LOOP: LDC1 F2, T3(T4)/* get T0[i+2]/T0[i+3] */ ADD.PS F1, F2, F8/* compute T1[i+2]/T1[i+3] */ ALNV.PS F3, F0, F1, T1/* align to dst memory */ SUXC1 F3, T3(T1)/* store to T1[i+0]/T1[i+1] */ ADDIU T3, 16 /* i = i + 4 */ LDC1 F2, T3(T0)/* get T0[i+0]/T0[i+1] */ ADD.PS F0, F2, F8/* compute T1[i+0]/T1[i+1] */ ALNV.PS F3, F1, F0, T1/* align to dst memory */ BNE T3, T2, LOOP SUXC1 F3, T3(T5)/* store to T1[i+2]/T1[i+3] */ /* Loop trailer stores all or half of F0, depending on T1 alignment */ The MIPS32® Instruction Set Manual, Revision 6.06 42 ALUIPC Aligned Add Upper Immediate to PC 31 26 25 21 20 16 15 0 655 16 Format: ALUIPC rs,immediate MIPS32 Release 6 Purpose: Aligned Add Upper Immediate to PC Description: GPR[rs]  ~0x0FFFF & ( PC + sign_extend( immediate << 16 ) ) This instruction performs a PC-relative address calculation. The 16-bit immediate is shifted left by 16 bits, sign- extended, and added to the address of the ALUIPC instruction. The low 16 bits of the result are cleared, that is the result is aligned on a 64K boundary. The result is placed in GPR rs. Restrictions: None Availability and Compatibility: This instruction is introduced by and required as of Release 6. Operation: GPR[rs]  ~0x0FFFF & ( PC + sign_extend( immediate << 16 ) ) Exceptions: None PCREL 111011 rs ALUIPC 11111 immediate The MIPS32® Instruction Set Manual, Revision 6.06 43 AND and 31 26 25 21 20 16 15 11 10 6 5 655556 Format: AND rd, rs, rt Purpose: and To do a bitwise logical AND. 0 MIPS32 SPECIAL 000000 rs rt rd 0 00000 AND 100100 44 The MIPS32® Instruction Set Manual, Revision 6.06 Description: GPR[rd]  GPR[rs] and GPR[rt] The contents of GPR rs are combined with the contents of GPR rt in a bitwise logical AND operation. The result is placed into GPR rd. Restrictions: None Operation: GPR[rd]  GPR[rs] and GPR[rt] Exceptions: None ANDI 31 26 25 21 20 16 15 655 16 Format: ANDI rt, rs, immediate Purpose: and immediate To do a bitwise logical AND with a constant Iand immediate 0 MIPS32 ANDI 001100 rs rt immediate Description: GPR[rt]  GPR[rs] and zero_extend(immediate) The 16-bit immediate is zero-extended to the left and combined with the contents of GPR rs in a bitwise logical AND operation. The result is placed into GPR rt. Restrictions: None Operation: GPR[rt]  GPR[rs] and zero_extend(immediate) Exceptions: None The MIPS32® Instruction Set Manual, Revision 6.06 45 ANDI and immediate 46 The MIPS32® Instruction Set Manual, Revision 6.06 AUI Add Immediate to Upper Bits 31 26 25 21 20 16 15 655 16 Format: AUI rt, rs immediate Purpose: Add Immediate to Upper Bits Add Upper Immediate Description: 0 MIPS32 Release 6 AUI 001111 rs rt immediate GPR[rt]  GPR[rs] + sign_extend(immediate << 16) The 16 bit immediate is shifted left 16 bits, sign-extended, and added to the register rs, storing the result in rt. In Release 6, LUI is an assembly idiom for AUI with rs=0. Restrictions: Availability and Compatibility: AUI is introduced by and required as of Release 6. Operation: GPR[rt]  GPR[rs] + sign_extend(immediate << 16) Exceptions: None. Programming Notes: AUI can be used to synthesize large constants in situations where it is not convenient to load a large constant from memory. To simplify hardware that may recognize sequences of instructions as generating large constants, AUI should be used in a stylized manner. To create an integer: LUI rd, imm_low(rtmp) ORI rd, rd, imm_upper To create a large offset for a memory access whose address is of the form rbase+large_offset: AUI rtmp, rbase, imm_upper LW rd, (rtmp)imm_low To create a large constant operand for an instruction of the form rd:=rs+large_immediate or rd:=rs-large_immediate: AUI rtmp,rs, imm_upper ADDIU rd, rtmp, imm_low The MIPS32® Instruction Set Manual, Revision 6.06 47 AUIPC Add Upper Immediate to PC 31 26 25 21 20 16 15 0 655 16 Format: AUIPC rs, immediate MIPS32 Release 6 Purpose: Add Upper Immediate to PC Description: GPR[rs]  ( PC + ( immediate << 16 ) ) This instruction performs a PC-relative address calculation. The 16-bit immediate is shifted left by 16 bits, sign- extended, and added to the address of the AUIPC instruction. The result is placed in GPR rs. Restrictions: None Availability and Compatibility: This instruction is introduced by and required as of Release 6. PCREL 111011 rs AUIPC 11110 immediate 48 The MIPS32® Instruction Set Manual, Revision 6.06 Operation: GPR[rs]  Exceptions: None ( PC + ( immediate << 16 ) ) B 31 26 25 21 20 16 15 655 16 Format: B offset Purpose: Unconditional Branch To do an unconditional branch. Description: branch IUnconditional Branch 0 Assembly Idiom BEQ 000100 0 00000 0 00000 offset B offset is the assembly idiom used to denote an unconditional branch. The actual instruction is interpreted by the hardware as BEQ r0, r0, offset. An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address. Restrictions: Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE. Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump. Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception. Operation: I: target_offset  sign_extend(offset || 02) I+1: PC  PC + target_offset Exceptions: None Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 Kbytes. Use jump (J) or jump register (JR) or the Release 6 branch compact (BC) instructions to branch to addresses outside this range. The MIPS32® Instruction Set Manual, Revision 6.06 49 BAL IBranch and Link pre-Release 6: 31 26 25 21 20 16 15 0 0 REGIMM 000001 00000 BGEZAL 10001 offset Release 6: 31 26 25 21 20 16 15 655 16 655 16 REGIMM 000001 0 00000 BAL 10001 offset Format: BALoffset AssemblyIdiomMIPS32,MIPS32Release6 Purpose: Branch and Link To do an unconditional PC-relative procedure call. Description: procedure_call Place the return address link in GPR 31. The return link is the address of the second instruction following the branch, where execution continues after a procedure call. An 18-bit signed offset (the 16-bit offset field shifted left 2-bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address. Restrictions: Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE. Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump. Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception. Availability and Compatibility: Pre-Release 6: BAL offset is the assembly idiom used to denote an unconditional branch. The actual instruction is interpreted by the hardware as BGEZAL r0, offset. Release 6 keeps the BAL special case of BGEZAL, but removes all other instances of BGEZAL. BGEZAL with rs any register other than GPR[0] is required to signal a Reserved Instruction exception. Operation: I: target_offset  sign_extend(offset || 02) GPR[31]  PC + 8 I+1: PC  PC + target_offset Exceptions: None Programming Notes: BAL without a corresponding return should NOT be used to read the PC. Doing so is likely to cause a performance loss on processors with a return address predictor. The MIPS32® Instruction Set Manual, Revision 6.06 50 BAL Branch and Link With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump and link (JAL) or jump and link register (JALR) instructions for procedure calls to addresses outside this range. 51 The MIPS32® Instruction Set Manual, Revision 6.06 BALC 31 IBranch and Link, Compact 0 MIPS32 Release 6 26 25 6 26 Format: BALC offset Purpose: Branch and Link, Compact To do an unconditional PC-relative procedure call. BALC 111010 offset Description: procedure_call (no delay slot) Place the return address link in GPR 31. The return link is the address of the instruction immediately following the branch, where execution continues after a procedure call. (Because compact branches have no delay slots, see below.) A 28-bit signed offset (the 26-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), to form a PC-relative effective target address. Compact branches do not have delay slots. The instruction after the branch is NOT executed when the branch is taken. Restrictions: This instruction is an unconditional, always taken, compact branch. It does not have a forbidden slot, that is, a Reserved Instruction exception is not caused by a Control Transfer Instruction placed in the slot following the branch. Availability and Compatibility: This instruction is introduced by and required as of Release 6. Release 6 instruction BALC occupies the same encoding as pre-Release 6 instruction SWC2. The SWC2 instruction has been moved to the COP2 major opcode in MIPS Release 6. Exceptions: None Operation: target_offset  sign_extend( offset || 02 ) GPR[31]  PC+4 PC  PC+4 + sign_extend(target_offset) The MIPS32® Instruction Set Manual, Revision 6.06 52 BC Branch, Compact 31 26 25 6 26 Format: BC offset Purpose: Branch, Compact 0 MIPS32 Release 6 BC 110010 offset 53 The MIPS32® Instruction Set Manual, Revision 6.06 Description: PC  PC+4 + sign_extend( offset << 2) A 28-bit signed offset (the 26-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), to form a PC-relative effective target address. Compact branches have no delay slot: the instruction after the branch is NOT executed when the branch is taken. Restrictions: This instruction is an unconditional, always taken, compact branch. It does not have a forbidden slot, that is, a Reserved Instruction exception is not caused by a Control Transfer Instruction placed in the slot following the branch. Availability and Compatibility: This instruction is introduced by and required as of Release 6. Release 6 instruction BC occupies the same encoding as pre-Release 6 instruction LWC2. The LWC2 instruction has been moved to the COP2 major opcode in MIPS Release 6. Exceptions: None Operation: target_offset  sign_extend( offset || 02 ) PC  ( PC+4 + sign_extend(target_offset) ) BC1EQZ BC1NEZ IBranch if Coprocessor 1 (FPU) Register Bit 0 Equal/Not Equal to Zero 31 26 25 21 20 16 15 655 16 Format: BC1EQZ BC1NEZ BC1EQZ ft, offset BC1NEZ ft, offset Purpose: Branch if Coprocessor 1 (FPU) Register Bit 0 Equal/Not Equal to Zero BC1EQZ: Branch if Coprocessor 1 (FPU) Register Bit 0 is Equal to Zero BC1NEZ: Branch if Coprocessor 1 (FPR) Register Bit 0 is Not Equal to Zero Description: BC1EQZ: if FPR[ft] & 1 = 0 then branch BC1NEZ: if FPR[ft] & 1  0 then branch The condition is evaluated on FPU register ft. • For BC1EQZ, the condition is true if and only if bit 0 of the FPU register ft is zero. • For BC1NEZ, the condition is true if and only if bit 0 of the FPU register ft is non-zero. If the condition is false, the branch is not taken, and execution continues with the next instruction. 0 MIPS32 Release 6 MIPS32 Release 6 COP1 010001 BC1EQZ 01001 ft offset COP1 010001 BC1NEZ 01101 ft offset 1. A 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), to form a PC-relative effective target address. Execute the instruction in the delay slot before the instruction at the target. Restrictions: If access to Coprocessor 1 is not enabled, a Coprocessor Unusable Exception is signaled. Because these instructions BC1EQZ and BC1NEZ do not depend on a particular floating point data type, they operate whenever Coprocessor 1 is enabled. Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE. If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 implementations are required to signal a Reserved Instruction exception. Availability and Compatibility: These instructions are introduced by and required as of Release 6. Exceptions: Coprocessor Unusable1 Operation: In Release 6, BC1EQZ and BC1NEZ are required, if the FPU is implemented. They must not signal a Reserved Instruction exception. They can signal a Coprocessor Unusable Exception. The MIPS32® Instruction Set Manual, Revision 6.06 54 55 The MIPS32® Instruction Set Manual, Revision 6.06 BC1EQZ BC1NEZ Branch if Coprocessor 1 (FPU) Register Bit 0 Equal/Not Equal to Zero tmp  ValueFPR(ft, UNINTERPRETED_WORD) BC1EQZ: cond  tmp & 1 = 0 BC1NEZ: cond  tmp & 1  0 if cond then I: target_PC  ( PC+4 + sign_extend( offset << 2 ) I+1: PC target_PC Programming Notes: Release 6: These instructions, BC1EQZ and BC1NEZ, replace the pre-Release 6 instructions BC1F and BC1T. These Release 6 FPU branches depend on bit 0 of the scalar FPU register. Note: BC1EQZ and BC1NEZ do not have a format or data type width. The same instructions are used for branches based on conditions involving any format, including 32-bit S (single precision) and W (word) format, and 64-bit D (double precision) and L (longword) format, as well as 128-bit MSA. The FPU scalar comparison instructions CMP.condn fmt produce an all ones or all zeros truth mask of their format width with the upper bits (where applica- ble) UNPREDICTABLE. BC1EQZ and BC1NEZ consume only bit 0 of the CMP.condn fmt output value, and there- fore operate correctly independent of fmt. BC1F 31 26 25 21 20 18 17 16 15 65311 16 Format: BC1F offset (cc = 0 implied) BC1F cc, offset Purpose: Branch on FP False To test an FP condition code and do a PC-relative conditional branch. Description: if FPConditionCode(cc) = 0 then branch IBranch on FP False 0 MIPS32, removed in Release 6 MIPS32, removed in Release 6 COP1 010001 BC 01000 cc nd 0 tf 0 offset An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself) in the branch delay slot to form a PC-relative effective target address. If the FP con- dition code bit cc is false (0), the program branches to the effective target address after the instruction in the delay slot is executed. An FP condition code is set by the FP compare instruction, C.cond fmt. Restrictions: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump. Availability and Compatibility: This instruction has been removed in Release 6. Operation: I: condition  FPConditionCode(cc) = 0 target_offset  (offset )GPRLEN-(16+2) || offset || 02 I+1: if condition then PC  PC + target_offset endif Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Unimplemented Operation Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. This instruction has been removed in Release 6 and has been replaced by the BC1EQZ instruction. Refer to the ‘BC1EQZ’ instruction in this manual for more information. Historical Information: The MIPS I architecture defines a single floating point condition code, implemented as the coprocessor 1 condition signal (Cp1Cond) and the C bit in the FP Control/Status register. MIPS I, II, and III architectures must have the CC field set to 0, which is implied by the first format in the “Format” section. The MIPS IV and MIPS32 architectures add seven more Condition Code bits to the original condition code 0. FP compare and conditional branch instructions specify the Condition Code bit to set or test. Both assembler formats are 15 The MIPS32® Instruction Set Manual, Revision 6.06 56 57 The MIPS32® Instruction Set Manual, Revision 6.06 BC1F Branch on FP False valid for MIPS IV and MIPS32. BC1FL IBranch on FP False Likely 31 26 25 21 20 18 17 16 15 65311 16 Format: BC1FL offset (cc = 0 implied) BC1FL cc, offset 0 MIPS32, removed in Release 6 MIPS32, removed in Release 6 COP1 010001 BC 01000 cc nd 1 tf 0 offset Purpose: Branch on FP False Likely To test an FP condition code and make a PC-relative conditional branch; execute the instruction in the delay slot only if the branch is taken. Description: if FPConditionCode(cc) = 0 then branch_likely An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself) in the branch delay slot to form a PC-relative effective target address. If the FP Con- dition Code bit cc is false (0), the program branches to the effective target address after the instruction in the delay slot is executed. If the branch is not taken, the instruction in the delay slot is not executed. An FP condition code is set by the FP compare instruction, C.cond fmt. Restrictions: Processor operation is UNPREDICTABLE if a branch, jump, ERET, DERET, or WAIT instruction is placed in the delay slot of a branch or jump. Availability and Compatibility: This instruction has been removed in Release 6. Operation: This operation specification is for the general Branch On Condition operation with the tf (true/false) and nd (nullify delay slot) fields as variables. The individual instructions BC1F, BC1FL, BC1T, and BC1TL have specific values for tf and nd. I: condition  FPConditionCode(cc) = 0 target_offset  (offset )GPRLEN-(16+2) || offset || 02 I+1: if condition then PC  PC + target_offset else NullifyCurrentInstruction() endif Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Unimplemented Operation Implementation Note: Some implementations always predict that the branch will be taken, and do not use nor do they update the branch internal processor branch prediction tables for this instruction. To maintain performance compatibility, future imple- mentations are encouraged to do the same. 15 The MIPS32® Instruction Set Manual, Revision 6.06 58 59 The MIPS32® Instruction Set Manual, Revision 6.06 BC1FL Branch on FP False Likely Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. In Pre-Release 6 implementations, software is strongly encouraged to avoid the use of the Branch Likely instructions, as they will be removed from a future revision of the MIPS Architecture. Some implementations always predict the branch will be taken, so there is a significant penalty if the branch is not taken. Software should only use this instruction when there is a very high probability (98% or more) that the branch will be taken. If the branch is not likely to be taken or if the probability of a taken branch is unknown, software is encouraged to use the BC1F instruction instead. Historical Information: The MIPS I architecture defines a single floating point condition code, implemented as the coprocessor 1 condition signal (Cp1Cond) and the C bit in the FP Control/Status register. MIPS I, II, and III architectures must have the CC field set to 0, which is implied by the first format in the “Format” section. The MIPS IV and MIPS32 architectures add seven more Condition Code bits to the original condition code 0. FP compare and conditional branch instructions specify the Condition Code bit to set or test. Both assembler formats are valid for MIPS IV and MIPS32. BC1T 31 26 25 21 20 18 17 16 15 65311 16 Format: BC1T offset (cc = 0 implied) BC1T cc, offset Purpose: Branch on FP True To test an FP condition code and do a PC-relative conditional branch. Description: if FPConditionCode(cc) = 1 then branch IBranch on FP True 0 MIPS32, removed in Release 6 MIPS32, removed in Release 6 COP1 010001 BC 01000 cc nd 0 tf 1 offset An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself) in the branch delay slot to form a PC-relative effective target address. If the FP con- dition code bit cc is true (1), the program branches to the effective target address after the instruction in the delay slot is executed. An FP condition code is set by the FP compare instruction, C.cond fmt. Restrictions: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump. Availability and Compatibility: This instruction has been removed in Release 6. Operation: I: condition  FPConditionCode(cc) = 1 target_offset  (offset )GPRLEN-(16+2) || offset || 02 I+1: if condition then PC  PC + target_offset endif Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Unimplemented Operation Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. This instruction has been replaced by the BC1NEZ instruction. Refer to the ‘BC1NEZ’ instruction in this manual for more information. Historical Information: The MIPS I architecture defines a single floating point condition code, implemented as the coprocessor 1 condition signal (Cp1Cond) and the C bit in the FP Control/Status register. MIPS I, II, and III architectures must have the CC field set to 0, which is implied by the first format in the “Format” section. 15 The MIPS32® Instruction Set Manual, Revision 6.06 60 61 The MIPS32® Instruction Set Manual, Revision 6.06 BC1T Branch on FP True The MIPS IV and MIPS32 architectures add seven more Condition Code bits to the original condition code 0. FP compare and conditional branch instructions specify the Condition Code bit to set or test. Both assembler formats are valid for MIPS IV and MIPS32. BC1TL 31 IBranch on FP True Likely 0 MIPS32, removed in Release 6 MIPS32, removed in Release 6 26 25 21 20 18 17 16 15 65311 16 Format: BC1TL offset (cc = 0 implied) BC1TL cc, offset COP1 010001 BC 01000 cc nd 1 tf 1 offset Purpose: Branch on FP True Likely To test an FP condition code and do a PC-relative conditional branch; execute the instruction in the delay slot only if the branch is taken. Description: if FPConditionCode(cc) = 1 then branch_likely An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself) in the branch delay slot to form a PC-relative effective target address. If the FP Con- dition Code bit cc is true (1), the program branches to the effective target address after the instruction in the delay slot is executed. If the branch is not taken, the instruction in the delay slot is not executed. An FP condition code is set by the FP compare instruction, C.cond fmt. Restrictions: Processor operation is UNPREDICTABLE if a branch, jump, ERET, DERET, or WAIT instruction is placed in the delay slot of a branch or jump. Availability and Compatibility: This instruction has been removed in Release 6. Operation: This operation specification is for the general Branch On Condition operation with the tf (true/false) and nd (nullify delay slot) fields as variables. The individual instructions BC1F, BC1FL, BC1T, and BC1TL have specific values for tf and nd. I: condition  FPConditionCode(cc) = 1 target_offset  (offset )GPRLEN-(16+2) || offset || 02 I+1: if condition then PC  PC + target_offset else NullifyCurrentInstruction() endif Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Unimplemented Operation Implementation Note: Some implementations always predict that the branch will be taken, and do not use nor do they update the branch internal processor branch prediction tables for this instruction. To maintain performance compatibility, future imple- mentations are encouraged to do the same. 15 The MIPS32® Instruction Set Manual, Revision 6.06 62 63 The MIPS32® Instruction Set Manual, Revision 6.06 BC1TL Branch on FP True Likely Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. In Pre-Release 6 implementations, software is strongly encouraged to avoid the use of the Branch Likely instructions, as they will be removed from a future revision of the MIPS Architecture. Some implementations always predict the branch will be taken, so there is a significant penalty if the branch is not taken. Software should only use this instruction when there is a very high probability (98% or more) that the branch will be taken. If the branch is not likely to be taken or if the probability of a taken branch is unknown, software is encouraged to use the BC1T instruction instead. Historical Information: The MIPS I architecture defines a single floating point condition code, implemented as the coprocessor 1 condition signal (Cp1Cond) and the C bit in the FP Control/Status register. MIPS I, II, and III architectures must have the CC field set to 0, which is implied by the first format in the “Format” section. The MIPS IV and MIPS32 architectures add seven more Condition Code bits to the original condition code 0. FP compare and conditional branch instructions specify the Condition Code bit to set or test. Both assembler formats are valid for MIPS IV and MIPS32. BC2EQZ BC2NEZ IBranch if Coprocessor 2 Condition (Register) Equal/Not Equal to Zero 31 26 25 21 20 16 15 655 16 Format: BC2EQZ BC2NEZ BC2EQZ ct, offset BC2NEZ ct, offset Purpose: Branch if Coprocessor 2 Condition (Register) Equal/Not Equal to Zero BC2EQZ: Branch if Coprocessor 2 Condition (Register) is Equal to Zero BC2NEZ: Branch if Coprocessor 2 Condition (Register) is Not Equal to Zero Description: BC2EQZ: if COP2Condition[ct] = 0 then branch BC2NEZ: if COP2Condition[ct]  0 then branch The 5-bit field ct specifies a coprocessor 2 condition. • For BC2EQZ if the coprocessor 2 condition is true the branch is taken. • For BC2NEZ if the coprocessor 2 condition is false the branch is taken. 0 MIPS32 Release 6 MIPS32 Release 6 COP2 010010 BC2EQZ 01001 ct offset COP2 010010 BC2NEZ 01101 ct offset A 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), to form a PC-relative effective target address. Execute the instruction in the delay slot before the instruction at the target. Restrictions: Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE. If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 implementations are required to signal a Reserved Instruction exception. If access to Coprocessor 2 is not enabled, a Coprocessor Unusable Exception is signaled. Availability and Compatibility: These instructions are introduced by and required as of Release 6. Exceptions: Coprocessor Unusable, Reserved Instruction Operation: tmpcond  Coprocessor2Condition(ct) if BC2EQZ then tmpcond  not(tmpcond) endif if tmpcond then PC  PC+4 + sign_extend( immediate << 2 ) ) endif The MIPS32® Instruction Set Manual, Revision 6.06 64 65 The MIPS32® Instruction Set Manual, Revision 6.06 BC2EQZ BC2NEZ Branch if Coprocessor 2 Condition (Register) Equal/Not Equal to Zero Implementation Notes: As of Release 6 these instructions, BC2EQZ and BC2NEZ, replace the pre-Release 6 instructions BC2F and BC2T, which had a 3-bit condition code field (as well as nullify and true/false bits). Release 6 makes all 5 bits of the ct con- dition code available to the coprocessor designer as a condition specifier. A customer defined coprocessor instruction set can implement any sort of condition it wants. For example, it could implement up to 32 single-bit flags, specified by the 5-bit field ct. It could also implement conditions encoded as values in a coprocessor register (such as testing the least significant bit of a coprocessor register) as done by Release 6 instructions BC1EQZ/BC1NEZ. BC2F 31 26 25 21 20 18 17 16 15 65311 16 Format: BC2F offset (cc = 0 implied) BC2F cc, offset Purpose: Branch on COP2 False To test a COP2 condition code and do a PC-relative conditional branch. Description: if COP2Condition(cc) = 0 then branch IBranch on COP2 False 0 MIPS32, removed in Release 6 MIPS32, removed in Release 6 COP2 010010 BC 01000 cc nd 0 tf 0 offset An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself) in the branch delay slot to form a PC-relative effective target address. If the COP2 condition specified by cc is false (0), the program branches to the effective target address after the instruction in the delay slot is executed. Restrictions: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump. Availability and Compatibility: This instruction has been removed in Release 6. Operation: I: condition  COP2Condition(cc) = 0 target_offset  (offset )GPRLEN-(16+2) || offset || 02 I+1: if condition then PC  PC + target_offset endif Exceptions: Coprocessor Unusable, Reserved Instruction Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. This instruction has been replaced by the BC2EQZ instruction. Refer to the ‘BC2EQZ’ instruction in this manual for more information. 15 The MIPS32® Instruction Set Manual, Revision 6.06 66 BC2FL Branch on COP2 False Likely 31 26 25 21 20 18 17 16 15 65311 16 Format: BC2FL offset (cc = 0 implied) BC2FL cc, offset 0 MIPS32, removed in Release 6 MIPS32, removed in Release 6 COP2 010010 BC 01000 cc nd 1 tf 0 offset 67 The MIPS32® Instruction Set Manual, Revision 6.06 Purpose: Branch on COP2 False Likely To test a COP2 condition code and make a PC-relative conditional branch; execute the instruction in the delay slot only if the branch is taken. Description: if COP2Condition(cc) = 0 then branch_likely An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself) in the branch delay slot to form a PC-relative effective target address. If the COP2 condition specified by cc is false (0), the program branches to the effective target address after the instruction in the delay slot is executed. If the branch is not taken, the instruction in the delay slot is not executed. Restrictions: Processor operation is UNPREDICTABLE if a branch, jump, ERET, DERET, or WAIT instruction is placed in the delay slot of a branch or jump. Availability and Compatibility: This instruction has been removed in Release 6. Operation: This operation specification is for the general Branch On Condition operation with the tf (true/false) and nd (nullify delay slot) fields as variables. The individual instructions BC2F, BC2FL, BC2T, and BC2TL have specific values for tf and nd. I: condition  COP2Condition(cc) = 0 target_offset  (offset )GPRLEN-(16+2) || offset || 02 I+1: if condition then PC  PC + target_offset else NullifyCurrentInstruction() endif Exceptions: Coprocessor Unusable, Reserved Instruction Implementation Note: Some implementations always predict that the branch will be taken, and do not use nor do they update the branch internal processor branch prediction tables for this instruction. To maintain performance compatibility, future imple- mentations are encouraged to do the same. Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. In Pre-Release 6 implementations, software is strongly encouraged to avoid the use of the Branch Likely instructions, 15 BC2FL IBranch on COP2 False Likely as they will be removed from a future revision of the MIPS Architecture. Some implementations always predict the branch will be taken, so there is a significant penalty if the branch is not taken. Software should only use this instruction when there is a very high probability (98% or more) that the branch will be taken. If the branch is not likely to be taken or if the probability of a taken branch is unknown, software is encouraged to use the BC2F instruction instead. The MIPS32® Instruction Set Manual, Revision 6.06 68 BC2T Branch on COP2 True 31 26 25 21 20 18 17 16 15 65311 16 Format: BC2T offset (cc = 0 implied) BC2T cc, offset Purpose: Branch on COP2 True To test a COP2 condition code and do a PC-relative conditional branch. Description: if COP2Condition(cc) = 1 then branch 0 MIPS32, removed in Release 6 MIPS32, removed in Release 6 COP2 010010 BC 01000 cc nd 0 tf 1 offset 69 The MIPS32® Instruction Set Manual, Revision 6.06 An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself) in the branch delay slot to form a PC-relative effective target address. If the COP2 condition specified by cc is true (1), the program branches to the effective target address after the instruction in the delay slot is executed. Restrictions: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump. Availability and Compatibility: This instruction has been removed in Release 6. Operation: I: condition  COP2Condition(cc) = 1 target_offset  (offset )GPRLEN-(16+2) || offset || 02 I+1: if condition then PC  PC + target_offset endif Exceptions: Coprocessor Unusable, Reserved Instruction Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. This instruction has been replaced by the BC2NEZ instruction. Refer to the ‘BC2NEZ’ instruction in this manual for more information. 15 BC2TL IBranch on COP2 True Likely 31 26 25 21 20 18 17 16 15 65311 16 Format: BC2TL offset (cc = 0 implied) BC2TL cc, offset 0 MIPS32, removed in Release 6 MIPS32, removed in Release 6 COP2 010010 BC 01000 cc nd 1 tf 1 offset Purpose: Branch on COP2 True Likely To test a COP2 condition code and do a PC-relative conditional branch; execute the instruction in the delay slot only if the branch is taken. Description: if COP2Condition(cc) = 1 then branch_likely An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself) in the branch delay slot to form a PC-relative effective target address. If the COP2 condition specified by cc is true (1), the program branches to the effective target address after the instruction in the delay slot is executed. If the branch is not taken, the instruction in the delay slot is not executed. Restrictions: Processor operation is UNPREDICTABLE if a branch, jump, ERET, DERET, or WAIT instruction is placed in the delay slot of a branch or jump. Availability and Compatibility: This instruction has been removed in Release 6. Operation: This operation specification is for the general Branch On Condition operation with the tf (true/false) and nd (nullify delay slot) fields as variables. The individual instructions BC2F, BC2FL, BC2T, and BC2TL have specific values for tf and nd. I: condition  COP2Condition(cc) = 1 target_offset  (offset )GPRLEN-(16+2) || offset || 02 I+1: if condition then PC  PC + target_offset else NullifyCurrentInstruction() endif Exceptions: Coprocessor Unusable, Reserved Instruction Implementation Note: Some implementations always predict that the branch will be taken, and do not use nor do they update the branch internal processor branch prediction tables for this instruction. To maintain performance compatibility, future imple- mentations are encouraged to do the same. Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. In Pre-Release 6 implementations, software is strongly encouraged to avoid the use of the Branch Likely instructions, 15 The MIPS32® Instruction Set Manual, Revision 6.06 70 71 The MIPS32® Instruction Set Manual, Revision 6.06 BC2TL Branch on COP2 True Likely as they will be removed from a future revision of the MIPS Architecture. Some implementations always predict the branch will be taken, so there is a significant penalty if the branch is not taken. Software should only use this instruction when there is a very high probability (98% or more) that the branch will be taken. If the branch is not likely to be taken or if the probability of a taken branch is unknown, software is encouraged to use the BC2T instruction instead. BEQ 31 26 25 21 20 16 15 655 16 Format: BEQ rs, rt, offset Purpose: Branch on Equal To compare GPRs then do a PC-relative conditional branch. IBranch on Equal 0 MIPS32 BEQ 000100 rs rt offset Description: if GPR[rs] = GPR[rt] then branch An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address. If the contents of GPR rs and GPR rt are equal, branch to the effective target address after the instruction in the delay slot is executed. Restrictions: Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE. Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump. Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception. Operation: I: target_offset  sign_extend(offset || 02) condition  (GPR[rs] = GPR[rt]) I+1: if condition then PC  PC + target_offset endif Exceptions: None Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. BEQ r0, r0 offset, expressed as B offset, is the assembly idiom used to denote an unconditional branch. The MIPS32® Instruction Set Manual, Revision 6.06 72 BEQL Branch on Equal Likely 31 26 25 21 20 16 15 655 16 Format: BEQL rs, rt, offset Purpose: Branch on Equal Likely 0 MIPS32, removed in Release 6 BEQL 010100 rs rt offset 73 The MIPS32® Instruction Set Manual, Revision 6.06 To compare GPRs then do a PC-relative conditional branch; execute the delay slot only if the branch is taken. Description: if GPR[rs] = GPR[rt] then branch_likely An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address. If the contents of GPR rs and GPR rt are equal, branch to the target address after the instruction in the delay slot is executed. If the branch is not taken, the instruction in the delay slot is not executed. Restrictions: Processor operation is UNPREDICTABLE if a branch, jump, ERET, DERET, or WAIT instruction is placed in the delay slot of a branch or jump. Availability and Compatibility: This instruction has been removed in Release 6. Operation: I: target_offset  sign_extend(offset || 02) condition  (GPR[rs] = GPR[rt]) I+1: if condition then PC  PC + target_offset else NullifyCurrentInstruction() endif Exceptions: None Implementation Note: Some implementations always predict that the branch will be taken, and do not use nor do they update the branch internal processor branch prediction tables for this instruction. To maintain performance compatibility, future imple- mentations are encouraged to do the same. Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. In Pre-Release 6 implementations, software is strongly encouraged to avoid the use of the Branch Likely instructions, as they will be removed from a future revision of the MIPS Architecture. Some implementations always predict the branch will be taken, so there is a significant penalty if the branch is not taken. Software should only use this instruction when there is a very high probability (98% or more) that the branch will be taken. If the branch is not likely to be taken or if the probability of a taken branch is unknown, software is encouraged to use the BEQ instruction instead. BEQL IBranch on Equal Likely Historical Information: In the MIPS I architecture, this instruction signaled a Reserved Instruction exception. The MIPS32® Instruction Set Manual, Revision 6.06 74 BGEZ Branch on Greater Than or Equal to Zero 31 26 25 21 20 16 15 0 655 16 Format: BGEZ rs, offset MIPS32 Purpose: Branch on Greater Than or Equal to Zero To test a GPR then do a PC-relative conditional branch Description: if GPR[rs]  0 then branch An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address. If the contents of GPR rs are greater than or equal to zero (sign bit is 0), branch to the effective target address after the instruction in the delay slot is executed. Restrictions: Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE. Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump. Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception. Operation: I: target_offset  sign_extend(offset || 02) condition  GPR[rs]  0GPRLEN I+1: if condition then PC  PC + target_offset endif Exceptions: None Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. REGIMM 000001 rs BGEZ 00001 offset 75 The MIPS32® Instruction Set Manual, Revision 6.06 BGEZAL IBranch on Greater Than or Equal to Zero and Link 31 26 25 21 20 16 15 655 16 Format: BGEZAL rs, offset Purpose: Branch on Greater Than or Equal to Zero and Link To test a GPR then do a PC-relative conditional procedure call 0 MIPS32, removed in Release 6 REGIMM 000001 rs BGEZAL 10001 offset Description: if GPR[rs]  0 then procedure_call Place the return address link in GPR 31. The return link is the address of the second instruction following the branch, where execution continues after a procedure call. An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address. If the contents of GPR rs are greater than or equal to zero (sign bit is 0), branch to the effective target address after the instruction in the delay slot is executed. Availability and Compatibility This instruction has been removed in Release 6 with the exception of special case BAL (unconditional Branch and Link) which was an alias for BGEZAL with rs=0. Restrictions: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump. Branch-and-link Restartability: GPR 31 must not be used for the source register rs, because such an instruction does not have the same effect when reexecuted. The result of executing such an instruction is UNPREDICTABLE. This restriction permits an exception handler to resume execution by reexecuting the branch when an exception occurs in the branch delay slot or forbidden slot. Operation: I: target_offset  sign_extend(offset || 02) condition  GPR[rs]  0GPRLEN GPR[31]  PC + 8 I+1: if condition then PC  PC + target_offset endif Exceptions: None Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump and link (JAL) or jump and link register (JALR) instructions for procedure calls to addresses outside this range. BGEZAL r0, offset, expressed as BAL offset, is the assembly idiom used to denote a PC-relative branch and link. BAL is used in a manner similar to JAL, but provides PC-relative addressing and a more limited target PC range. The MIPS32® Instruction Set Manual, Revision 6.06 76 B{LE,GE,GT,LT,EQ,NE}ZALC ICompact Zero-Compare and Branch-and-Link Instructions 31 26 25 21 20 16 15 0 POP06 000110 BLEZALC 00000 rt  00000 offset POP06 000110 BGEZALC rs = rt  00000 rs rt offset POP07 000111 BGTZALC 00000 rt  00000 offset POP07 000111 BLTZALC rs = rt  00000 rs rt offset POP10 001000 00000 BEQZALC rs < rt rt  00000 offset POP30 011000 00000 BNEZALC rs < rt rt  00000 offset Format: B{LE,GE,GT,LT,EQ,NE}ZALC BLEZALC rt, offset BGEZALC rt, offset BGTZALC rt, offset BLTZALC rt, offset BEQZALC rt, offset BNEZALC rt, offset 655 16 Purpose: Compact Zero-Compare and Branch-and-Link Instructions BLEZALC: Compact branch-and-link if GPR rt is less than or equal to zero BGEZALC: Compact branch-and-link if GPR rt is greater than or equal to zero BGTZALC: Compact branch-and-link if GPR rt is greater than zero BLTZALC: Compact branch-and-link if GPR rt is less than to zero BEQZALC: Compact branch-and-link if GPR rt is equal to zero BNEZALC: Compact branch-and-link if GPR rt is not equal to zero Description: if condition(GPR[rt]) then procedure_call branch (no delay slot) The condition is evaluated. If the condition is true, the branch is taken. Places the return address link in GPR 31. The return link is the address of the instruction immediately following the branch, where execution continues after a procedure call. The return address link is unconditionally updated. A 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), to form a PC-relative effective target address. The MIPS32® Instruction Set Manual, Revision 6.06 77 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 78 The MIPS32® Instruction Set Manual, Revision 6.06 B{LE,GE,GT,LT,EQ,NE}ZALC Compact Zero-Compare and Branch-and-Link Instructions BLEZALC: the condition is true if and only if GPR rt is less than or equal to zero. BGEZALC: the condition is true if and only if GPR rt is greater than or equal to zero. BLTZALC: the condition is true if and only if GPR rt is less than zero. BGTZALC: the condition is true if and only if GPR rt is greater than zero. BEQZALC: the condition is true if and only if GPR rt is equal to zero. BNEZALC: the condition is true if and only if GPR rt is not equal to zero. Compact branches do not have delay slots. The instruction after a compact branch is only executed if the branch is not taken. Restrictions: Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE. If a control transfer instruction (CTI) is executed in the forbidden slot of a compact branch, Release 6 implementa- tions are required to signal a Reserved Instruction exception, but only when the branch is not taken. Branch-and-link Restartability: GPR 31 must not be used for the source registers, because such an instruction does not have the same effect when reexecuted. The result of executing such an instruction is UNPREDICTABLE. This restriction permits an exception handler to resume execution by reexecuting the branch when an exception occurs in the branch delay slot or forbidden slot. Availability and Compatibility: These instructions are introduced by and required as of Release 6. • BEQZALC reuses the opcode assigned to pre-Release 6 ADDI. • BNEZALC reuses the opcode assigned to pre-Release 6 MIPS64 DADDI. These instructions occupy primary opcode spaces originally allocated to other instructions. BLEZALC and BGEZALC have the same primary opcode as BLEZ, and are distinguished by rs and rt register numbers. Similarly, BGTZALC and BLTZALC have the same primary opcode as BGTZ, and are distinguished by register fields. BEQZALC and BNEZALC reuse the primary opcodes ADDI and DADDI. Exceptions: None Operation: GPR[31]  PC+4 target_offset  sign_extend( offset || 02 ) BLTZALC: cond  GPR[rt] < 0 BLEZALC: cond  GPR[rt]  0 BGEZALC: cond  GPR[rt]  0 BGTZALC: cond  GPR[rt] > 0 BEQZALC: cond  GPR[rt] = 0 BNEZALC: cond  GPR[rt]  0
if cond then
PC  ( PC+4+ sign_extend(
endif
Programming Notes:
target_offset ) )
Software that performs incomplete instruction decode may incorrectly decode these new instructions, because of their

B{LE,GE,GT,LT,EQ,NE}ZALC ICompact Zero-Compare and Branch-and-Link Instructions
very tight encoding. For example, a disassembler might look only at the primary opcode field, instruction bits 31-26, to decode BLEZL without checking that the “rt” field is zero. Such software violated the pre-Release 6 architecture specification.
With the 16-bit offset shifted left 2 bits and sign extended, the conditional branch range is ± 128 KBytes. Other instructions such as pre-Release 6 JAL and JALR, or Release 6 JIALC and BALC have larger ranges. In particular, BALC, with a 26-bit offset shifted by 2 bits, has a 28-bit range, ± 128 MBytes. Code sequences using AUIPC and JIALC allow still greater PC-relative range.
The MIPS32® Instruction Set Manual, Revision 6.06 79

BGEZALL IBranch on Greater Than or Equal to Zero and Link Likely
31
26 25 21 20 16 15 0
655 16
Format: BGEZALL rs, offset MIPS32, removed in Release 6 Purpose: Branch on Greater Than or Equal to Zero and Link Likely
To test a GPR then do a PC-relative conditional procedure call; execute the delay slot only if the branch is taken.
Description: if GPR[rs]  0 then procedure_call_likely
Place the return address link in GPR 31. The return link is the address of the second instruction following the branch,
where execution continues after a procedure call.
An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address.
If the contents of GPR rs are greater than or equal to zero (sign bit is 0), branch to the effective target address after the instruction in the delay slot is executed. If the branch is not taken, the instruction in the delay slot is not executed.
Restrictions:
Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a
branch or jump.
Branch-and-link Restartability: GPR 31 must not be used for the source register rs, because such an instruction does not have the same effect when reexecuted. The result of executing such an instruction is UNPREDICTABLE. This restriction permits an exception handler to resume execution by reexecuting the branch when an exception occurs in the branch delay slot.
Availability and Compatibility:
This instruction has been removed in Release 6.
Operation:
I: target_offset  sign_extend(offset || 02) condition  GPR[rs]  0GPRLEN
GPR[31]  PC + 8
I+1: if condition then
PC  PC + target_offset
else
NullifyCurrentInstruction()
endif
Exceptions:
None
Programming Notes:
With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump and link (JAL) or jump and link register (JALR) instructions for procedure calls to addresses outside this range.
Some implementations always predict the branch will be taken, so there is a significant penalty if the branch is not taken. Software should only use this instruction when there is a very high probability (98% or more) that the branch will be taken. If the branch is not likely to be taken or if the probability of a taken branch is unknown, software is
REGIMM 000001
rs
BGEZALL 10011
offset
The MIPS32® Instruction Set Manual, Revision 6.06 80

BGEZALL IBranch on Greater Than or Equal to Zero and Link Likely encouraged to use the BGEZAL instruction instead.
Historical Information:
In the MIPS I architecture, this instruction signaled a Reserved Instruction exception.
The MIPS32® Instruction Set Manual, Revision 6.06 81

BC ICompact Compare-and-Branch Instructions
31
26 25 21 20 16 15
0
POP26 010110
BLEZC
00000
rt  00000
offset
POP26 010110
BGEZC rs = rt
rs  00000 rt  00000
offset
POP26 010110
BGEC (BLEC) rs  rt
rs  00000 rt  00000
offset
POP27 010111
BGTZC
00000
rt  00000
offset
POP27 010111
BLTZC rs = rt
rs  00000 rt  00000
offset
POP27 010111
BLTC (BGTC) rs  rt
rs  00000 rt  00000
offset
POP06 000110
BGEUC (BLEUC) rs  rt
rs  00000 rt  00000
offset
POP07 000111
BLTUC (BGTUC) rs  rt
rs  00000 rt  00000
offset
POP10 001000
BEQC rs < rt rs  00000 rt  00000 offset POP30 011000 BNEC rs < rt rs  00000 rt  00000 offset 31 655 16 26 25 21 20 65 21 Format: BC rs, rt, offset
Purpose: Compact Compare-and-Branch Instructions Format Details:
Equal/Not-Equal register-register compare and branch with 16-bit offset:
BEQC rs, rt, offset
BNEC rs, rt, offset
0
MIPS32 Release 6
MIPS32 Release 6 MIPS32 Release 6
POP66 110110
BEQZC rs  00000 rs
offset
POP76 111110
BNEZC rs  00000 rs
offset
The MIPS32® Instruction Set Manual, Revision 6.06
82

83
The MIPS32® Instruction Set Manual, Revision 6.06
BC
Compact Compare-and-Branch Instructions
Signed register-register compare and branch with 16-bit offset:
BLTC rs, rt, offset
BGEC rs, rt, offset
Unsigned register-register compare and branch with 16-bit offset:
BLTUC rs, rt, offset
BGEUC rs, rt, offset
Assembly idioms with reversed operands for signed/unsigned compare-and-branch:
BGTC rt, rs, offset
BLEC rt, rs, offset
BGTUC rt, rs, offset
BLEUC rt, rs, offset
Signed Compare register to Zero and branch with 16-bit offset:
BLTZC rt, offset
BLEZC rt, offset
BGEZC rt, offset
BGTZC rt, offset
Equal/Not-equal Compare register to Zero and branch with 21-bit offset:
BEQZC rs, offset
BNEZC rs, offset
MIPS32 Release 6 MIPS32 Release 6
MIPS32 Release 6 MIPS32 Release 6
Assembly Idiom Assembly Idiom Assembly Idiom Assembly Idiom
MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6
MIPS32 Release 6 MIPS32 Release 6
Description: if condition(GPR[rs] and/or GPR[rt]) then compact branch (no delay slot) The condition is evaluated. If the condition is true, the branch is taken.
An 18/23-bit signed offset (the 16/21-bit offset field shifted left 2 bits) is added to the address of the instruction fol- lowing the branch (not the branch itself), to form a PC-relative effective target address.
The offset is 16 bits for most compact branches, including BLTC, BLEC, BGEC, BGTC, BNEQC, BNEC, BLTUC, BLEUC, BGEUC, BGTC, BLTZC, BLEZC, BGEZC, BGTZC. The offset is 21 bits for BEQZC and BNEZC.
Compact branches have no delay slot: the instruction after the branch is NOT executed if the branch is taken. The conditions are as follows:
Equal/Not-equal register-register compare-and-branch with 16-bit offset: BEQC: Compact branch if GPRs are equal
BNEC: Compact branch if GPRs are not equal
Signed register-register compare and branch with 16-bit offset:
BLTC: Compact branch if GPR rs is less than GPR rt
BGEC: Compact branch if GPR rs is greater than or equal to GPR rt
Unsigned register-register compare and branch with 16-bit offset:
BLTUC: Compact branch if GPR rs is less than GPR rt, unsigned
BGEUC: Compact branch if GPR rs is greater than or equal to GPR rt, unsigned
Assembly Idioms with Operands Reversed:
BLEC: Compact branch if GPR rt is less than or equal to GPR rs (alias for BGEC)
BGTC: Compact branch if GPR rt is greater than GPR rs (alias for BLTC)
BLEUC: Compact branch if GPR rt is less than or equal to GPR rt, unsigned (alias for BGEUC) BGTUC: Compact branch if GPR rt is greater than GPR rs, unsigned (alias for BLTUC)

BC ICompact Compare-and-Branch Instructions
Compare register to zero and branch with 16-bit offset:
BLTZC: Compact branch if GPR rt is less than zero
BLEZC: Compact branch if GPR rt is less than or equal to zero BGEZC: Compact branch if GPR rt is greater than or equal to zero BGTZC: Compact branch if GPR rt is greater than zero
Compare register to zero and branch with 21-bit offset:
BEQZC: Compact branch if GPR rs is equal to zero BNEZC: Compact branch if GPR rs is not equal to zero
Restrictions:
Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE.
If a control transfer instruction (CTI) is placed in the forbidden slot of a compact branch, Release 6 implementations are required to signal a Reserved Instruction exception, but only when the branch is not taken.
Availability and Compatibility:
These instructions are introduced by and required as of Release 6.
• BEQZC reuses the opcode assigned to pre-Release 6 LDC2.
• BNEZC reuses the opcode assigned to pre-Release 6 SDC2.
• BEQC reuses the opcode assigned to pre-Release 6 ADDI.
• BNEC reuses the opcode assigned to pre-Release 6 MIPD64 DADDI.
Exceptions:
None
Operation:
target_offset  sign_extend( offset || 02 )
/* Register-register compare and branch, 16 bit offset: */ /* Equal / Not-Equal */
BEQC: cond  GPR[rs] = GPR[rt]
BNEC: cond  GPR[rs]  GPR[rt]
/* Signed */
BLTC: cond  GPR[rs] < GPR[rt] BGEC: cond  GPR[rs]  GPR[rt] /* Unsigned: */ BLTUC: cond  unsigned(GPR[rs]) < unsigned(GPR[rt]) BGEUC: cond  unsigned(GPR[rs])  unsigned(GPR[rt]) /* Compare register to zero, small offset: */ BLTZC: cond  GPR[rt] < 0 BLEZC: cond  GPR[rt]  0 BGEZC: cond  GPR[rt]  0 BGTZC: cond  GPR[rt] > 0
/* Compare register to zero, large offset: */ BEQZC: cond  GPR[rs] = 0
BNEZC: cond  GPR[rs]  0
if cond then
PC  ( PC+4+ sign_extend( offset ) )
The MIPS32® Instruction Set Manual, Revision 6.06 84

85
The MIPS32® Instruction Set Manual, Revision 6.06
BC Compact Compare-and-Branch Instructions end if
Programming Notes:
Legacy software that performs incomplete instruction decode may incorrectly decode these new instructions, because of their very tight encoding. For example, a disassembler that looks only at the primary opcode field (instruction bits 31-26) to decode BLEZL without checking that the “rt” field is zero violates the pre-Release 6 architecture specifica- tion. Complete instruction decode allows reuse of pre-Release 6 BLEZL opcode for Release 6 conditional branches.

BGEZL IBranch on Greater Than or Equal to Zero Likely
31
26 25 21 20 16 15 0
655 16
Format: BGEZL rs, offset MIPS32, removed in Release 6 Purpose: Branch on Greater Than or Equal to Zero Likely
To test a GPR then do a PC-relative conditional branch; execute the delay slot only if the branch is taken.
Description: if GPR[rs]  0 then branch_likely
An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following
the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address.
If the contents of GPR rs are greater than or equal to zero (sign bit is 0), branch to the effective target address after the instruction in the delay slot is executed. If the branch is not taken, the instruction in the delay slot is not executed.
Restrictions:
Processor operation is UNPREDICTABLE if a branch, jump, ERET, DERET, or WAIT instruction is placed in the
delay slot of a branch or jump.
Availability and Compatibility:
This instruction has been removed in Release 6.
Operation:
I: target_offset  sign_extend(offset || 02) condition  GPR[rs]  0GPRLEN
I+1: if condition then
PC  PC + target_offset
else
NullifyCurrentInstruction()
endif
Exceptions:
None
Implementation Note:
Some implementations always predict that the branch will be taken, and do not use nor do they update the branch internal processor branch prediction tables for this instruction. To maintain performance compatibility, future imple- mentations are encouraged to do the same.
Programming Notes:
With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range.
In Pre-Release 6 implementations, software is strongly encouraged to avoid the use of the Branch Likely instructions, as they will be removed from a future revision of the MIPS Architecture.
Some implementations always predict the branch will be taken, so there is a significant penalty if the branch is not taken. Software should only use this instruction when there is a very high probability (98% or more) that the branch will be taken. If the branch is not likely to be taken or if the probability of a taken branch is unknown, software is encouraged to use the BGEZ instruction instead.
REGIMM 000001
rs
BGEZL 00011
offset
The MIPS32® Instruction Set Manual, Revision 6.06 86

BGEZL Branch on Greater Than or Equal to Zero Likely Historical Information:
In the MIPS I architecture, this instruction signaled a Reserved Instruction exception.
87 The MIPS32® Instruction Set Manual, Revision 6.06

BGTZ IBranch on Greater Than Zero
31
26 25 21 20 16 15 0
655 16
Format: BGTZ rs, offset MIPS32 Purpose: Branch on Greater Than Zero
To test a GPR then do a PC-relative conditional branch.
Description: if GPR[rs] > 0 then branch
An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following
the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address.
If the contents of GPR rs are greater than zero (sign bit is 0 but value not zero), branch to the effective target address after the instruction in the delay slot is executed.
Restrictions:
Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE.
Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump.
Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception.
Operation:
I: target_offset  sign_extend(offset || 02) condition  GPR[rs] > 0GPRLEN
I+1: if condition then
PC  PC + target_offset
endif
Exceptions:
None
Programming Notes:
With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range.
BGTZ 000111
rs
0 00000
offset
The MIPS32® Instruction Set Manual, Revision 6.06 88

BGTZL Branch on Greater Than Zero Likely
31
26 25 21 20 16 15 0
655 16
Format: BGTZL rs, offset MIPS32, removed in Release 6 Purpose: Branch on Greater Than Zero Likely
To test a GPR then do a PC-relative conditional branch; execute the delay slot only if the branch is taken.
Description: if GPR[rs] > 0 then branch_likely
An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following
the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address.
If the contents of GPR rs are greater than zero (sign bit is 0 but value not zero), branch to the effective target address after the instruction in the delay slot is executed. If the branch is not taken, the instruction in the delay slot is not exe- cuted.
Restrictions:
Processor operation is UNPREDICTABLE if a branch, jump, ERET, DERET, or WAIT instruction is placed in the
delay slot of a branch or jump.
Availability and Compatibility:
This instruction has been removed in Release 6.
Operation:
I: target_offset  sign_extend(offset || 02) condition  GPR[rs] > 0GPRLEN
I+1: if condition then
PC  PC + target_offset
else
NullifyCurrentInstruction()
endif
Exceptions:
None
Implementation Note:
Some implementations always predict that the branch will be taken, and do not use nor do they update the branch internal processor branch prediction tables for this instruction. To maintain performance compatibility, future imple- mentations are encouraged to do the same.
Programming Notes:
With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range.
In Pre-Release 6 implementations, software is strongly encouraged to avoid the use of the Branch Likely instructions, as they will be removed from a future revision of the MIPS Architecture.
Some implementations always predict the branch will be taken, so there is a significant penalty if the branch is not taken. Software should only use this instruction when there is a very high probability (98% or more) that the branch will be taken. If the branch is not likely to be taken or if the probability of a taken branch is unknown, software is
BGTZL 010111
rs
0 00000
offset
89
The MIPS32® Instruction Set Manual, Revision 6.06

BGTZL IBranch on Greater Than Zero Likely encouraged to use the BGTZ instruction instead.
Historical Information:
In the MIPS I architecture, this instruction signaled a Reserved Instruction exception.
The MIPS32® Instruction Set Manual, Revision 6.06 90

BGTZL Branch on Greater Than Zero Likely
91 The MIPS32® Instruction Set Manual, Revision 6.06

BITSWAP ISwaps (reverses) bits in each byte
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: BITSWAP
BITSWAP rd,rt MIPS32 Release 6
Purpose: Swaps (reverses) bits in each byte
Description: GPR[rd].byte(i)  reverse_bits_in_byte(GPR[rt].byte(i)), for all
bytes i
Each byte in input GPR rt is moved to the same byte position in output GPR rd, with bits in each byte reversed. BITSWAP operates on all 4 bytes of a 32-bit GPR on a 32-bit CPU.
Restrictions:
None.
Availability and Compatibility:
The BITSWAP instruction is introduced by and required as of Release 6.
Operation:
BITSWAP:
for i in 0 to 3 do /* for all bytes in 32-bit GPR width */
tmp.byte(i)  reverse_bits_in_byte( GPR[rt].byte(i) ) endfor
GPR[rd]  tmp where
function reverse_bits_in_byte(inbyte) outbyte7  inbyte0
outbyte6  inbyte1
outbyte5  inbyte2
outbyte4  inbyte3 outbyte3  inbyte4 outbyte2  inbyte5 outbyte1  inbyte6 outbyte0  inbyte7 return outbyte
end function
Exceptions:
None
Programming Notes:
The Release 6 BITSWAP instruction corresponds to the DSP Module BITREV instruction, except that the latter bit- reverses the least-significant 16-bit halfword of the input register, zero extending the rest, while BITSWAP operates on 32-bits.
SPECIAL3 011111
00000
rt
rd
BITSWAP 00000
BSHFL 100000
The MIPS32® Instruction Set Manual, Revision 6.06 92

BITSWAP Swaps (reverses) bits in each byte
93 The MIPS32® Instruction Set Manual, Revision 6.06

BLEZ
IBranch on Less Than or Equal to Zero
31
26 25 21 20 16 15 0
655 16
Format: BLEZ rs, offset MIPS32 Purpose: Branch on Less Than or Equal to Zero
To test a GPR then do a PC-relative conditional branch.
Description: if GPR[rs]  0 then branch
An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following
the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address.
If the contents of GPR rs are less than or equal to zero (sign bit is 1 or value is zero), branch to the effective target address after the instruction in the delay slot is executed.
Restrictions:
Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE.
Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump.
Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception.
Operation:
I: target_offset  sign_extend(offset || 02) condition  GPR[rs]  0GPRLEN
I+1: if condition then
PC  PC + target_offset
endif
Exceptions:
None
Programming Notes:
With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range.
BLEZ 000110
rs
0 00000
offset
The MIPS32® Instruction Set Manual, Revision 6.06 94

BLEZL Branch on Less Than or Equal to Zero Likely
31
26 25 21 20 16 15 0
655 16
Format: BLEZL rs, offset MIPS32, removed in Release 6 Purpose: Branch on Less Than or Equal to Zero Likely
To test a GPR then do a PC-relative conditional branch; execute the delay slot only if the branch is taken.
Description: if GPR[rs]  0 then branch_likely
An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following
the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address.
If the contents of GPR rs are less than or equal to zero (sign bit is 1 or value is zero), branch to the effective target address after the instruction in the delay slot is executed. If the branch is not taken, the instruction in the delay slot is not executed.
Restrictions:
Processor operation is UNPREDICTABLE if a branch, jump, ERET, DERET, or WAIT instruction is placed in the
delay slot of a branch or jump.
Availability and Compatibility:
This instruction has been removed in Release 6.
Operation:
I: target_offset  sign_extend(offset || 02) condition  GPR[rs]  0GPRLEN
I+1: if condition then
PC  PC + target_offset
else
NullifyCurrentInstruction()
endif
Exceptions:
None
Implementation Note:
Some implementations always predict that the branch will be taken, and do not use nor do they update the branch internal processor branch prediction tables for this instruction. To maintain performance compatibility, future imple- mentations are encouraged to do the same.
Programming Notes:
With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range.
In Pre-Release 6 implementations, software is strongly encouraged to avoid the use of the Branch Likely instructions, as they will be removed from a future revision of the MIPS Architecture.
Some implementations always predict the branch will be taken, so there is a significant penalty if the branch is not taken. Software should only use this instruction when there is a very high probability (98% or more) that the branch will be taken. If the branch is not likely to be taken or if the probability of a taken branch is unknown, software is
BLEZL 010110
rs
0 00000
offset
95
The MIPS32® Instruction Set Manual, Revision 6.06

BLEZL IBranch on Less Than or Equal to Zero Likely encouraged to use the BLEZ instruction instead.
Historical Information:
In the MIPS I architecture, this instruction signaled a Reserved Instruction exception.
The MIPS32® Instruction Set Manual, Revision 6.06 96

BLTZ
Branch on Less Than Zero
31
26 25 21 20 16 15 0
655 16
Format: BLTZ rs, offset MIPS32 Purpose: Branch on Less Than Zero
To test a GPR then do a PC-relative conditional branch.
Description: if GPR[rs] < 0 then branch An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address. If the contents of GPR rs are less than zero (sign bit is 1), branch to the effective target address after the instruction in the delay slot is executed. Restrictions: Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE. Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump. Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception. Operation: I: target_offset  sign_extend(offset || 02) condition  GPR[rs] < 0GPRLEN I+1: if condition then PC  PC + target_offset endif Exceptions: None Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump and link (JAL) or jump and link register (JALR) instructions for procedure calls to addresses outside this range. REGIMM 000001 rs BLTZ 00000 offset 97 The MIPS32® Instruction Set Manual, Revision 6.06 BLTZAL IBranch on Less Than Zero and Link 31 26 25 21 20 16 15 655 16 Format: BLTZAL rs, offset Purpose: Branch on Less Than Zero and Link To test a GPR then do a PC-relative conditional procedure call. 0 MIPS32, removed in Release 6 REGIMM 000001 rs BLTZAL 10000 offset Description: if GPR[rs] < 0 then procedure_call Place the return address link in GPR 31. The return link is the address of the second instruction following the branch, where execution continues after a procedure call. An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address. If the contents of GPR rs are less than zero (sign bit is 1), branch to the effective target address after the instruction in the delay slot is executed. Availability and Compatibility: This instruction has been removed in Release 6. The special case BLTZAL r0, offset, has been retained as NAL in Release 6. Restrictions: Processor operation is UNPREDICTABLE if a branch, jump, ERET, DERET, or WAIT instruction is placed in the delay slot of a branch or jump. Branch-and-link Restartability: GPR 31 must not be used for the source register rs, because such an instruction does not have the same effect when re-executed. The result of executing such an instruction is UNPREDICTABLE. This restriction permits an exception handler to resume execution by re-executing the branch when an exception occurs in the branch delay slot. Operation: I: target_offset  sign_extend(offset || 02) condition  GPR[rs] < 0GPRLEN GPR[31]  PC + 8 I+1: if condition then PC  PC + target_offset endif Exceptions: None Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump and link (JAL) or jump and link register (JALR) instructions for procedure calls to addresses outside this range. The MIPS32® Instruction Set Manual, Revision 6.06 98 BLTZALL Branch on Less Than Zero and Link Likely 31 26 25 21 20 16 15 0 655 16 Format: BLTZALL rs, offset MIPS32, removed in Release 6 Purpose: Branch on Less Than Zero and Link Likely To test a GPR then do a PC-relative conditional procedure call; execute the delay slot only if the branch is taken. Description: if GPR[rs] < 0 then procedure_call_likely Place the return address link in GPR 31. The return link is the address of the second instruction following the branch, where execution continues after a procedure call. An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address. If the contents of GPR rs are less than zero (sign bit is 1), branch to the effective target address after the instruction in the delay slot is executed. If the branch is not taken, the instruction in the delay slot is not executed. Restrictions: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump. Branch-and-link Restartability: GPR 31 must not be used for the source register rs, because such an instruction does not have the same effect when reexecuted. The result of executing such an instruction is UNPREDICTABLE. This restriction permits an exception handler to resume execution by reexecuting the branch when an exception occurs in the branch delay slot. Availability and Compatibility: This instruction has been removed in Release 6. Operation: I: target_offset  sign_extend(offset || 02) condition  GPR[rs] < 0GPRLEN GPR[31]  PC + 8 I+1: if condition then PC  PC + target_offset else NullifyCurrentInstruction() endif Exceptions: None Implementation Note: Some implementations always predict that the branch will be taken, and do not use nor do they update the branch internal processor branch prediction tables for this instruction. To maintain performance compatibility, future imple- mentations are encouraged to do the same. Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump and link (JAL) or The MIPS32® Instruction Set Manual, Revision 6.06 REGIMM 000001 rs BLTZALL 10010 offset 99 BLTZALL IBranch on Less Than Zero and Link Likely jump and link register (JALR) instructions for procedure calls to addresses outside this range. In Pre-Release 6 implementations, software is strongly encouraged to avoid the use of the Branch Likely instructions, as they will be removed from a future revision of the MIPS Architecture. Some implementations always predict the branch will be taken, so there is a significant penalty if the branch is not taken. Software should only use this instruction when there is a very high probability (98% or more) that the branch will be taken. If the branch is not likely to be taken or if the probability of a taken branch is unknown, software is encouraged to use the BLTZAL instruction instead. Historical Information: In the MIPS I architecture, this instruction signaled a Reserved Instruction exception. The MIPS32® Instruction Set Manual, Revision 6.06 100 BLTZL Branch on Less Than Zero Likely 31 26 25 21 20 16 15 0 655 16 Format: BLTZL rs, offset MIPS32, removed in Release 6 Purpose: Branch on Less Than Zero Likely To test a GPR then do a PC-relative conditional branch; execute the delay slot only if the branch is taken. Description: if GPR[rs] < 0 then branch_likely An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address. If the contents of GPR rs are less than zero (sign bit is 1), branch to the effective target address after the instruction in the delay slot is executed. If the branch is not taken, the instruction in the delay slot is not executed. Restrictions: Processor operation is UNPREDICTABLE if a branch, jump, ERET, DERET, or WAIT instruction is placed in the delay slot of a branch or jump. Availability and Compatibility: This instruction has been removed in Release 6. Operation: I: target_offset  sign_extend(offset || 02) condition  GPR[rs] < 0GPRLEN I+1: if condition then PC  PC + target_offset else NullifyCurrentInstruction() endif Exceptions: None Implementation Note: Some implementations always predict that the branch will be taken, and do not use nor do they update the branch internal processor branch prediction tables for this instruction. To maintain performance compatibility, future imple- mentations are encouraged to do the same. Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. In Pre-Release 6 implementations, software is strongly encouraged to avoid the use of the Branch Likely instructions, as they will be removed from a future revision of the MIPS Architecture. Some implementations always predict the branch will be taken, so there is a significant penalty if the branch is not taken. Software should only use this instruction when there is a very high probability (98% or more) that the branch will be taken. If the branch is not likely to be taken or if the probability of a taken branch is unknown, software is encouraged to use the BLTZ instruction instead. REGIMM 000001 rs BLTZL 00010 offset 101 The MIPS32® Instruction Set Manual, Revision 6.06 BLTZL IBranch on Less Than Zero Likely Historical Information: In the MIPS I architecture, this instruction signaled a Reserved Instruction exception. The MIPS32® Instruction Set Manual, Revision 6.06 102 BNE Branch on Not Equal 31 26 25 21 20 16 15 655 16 Format: BNE rs, rt, offset Purpose: Branch on Not Equal To compare GPRs then do a PC-relative conditional branch 0 MIPS32 BNE 000101 rs rt offset 103 The MIPS32® Instruction Set Manual, Revision 6.06 Description: if GPR[rs]  GPR[rt] then branch An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address. If the contents of GPR rs and GPR rt are not equal, branch to the effective target address after the instruction in the delay slot is executed. Restrictions: Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE. Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump. Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception. Operation: I: target_offset  sign_extend(offset || 02) condition  (GPR[rs]  GPR[rt]) I+1: if condition then PC  PC + target_offset endif Exceptions: None Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. BNEL IBranch on Not Equal Likely 31 26 25 21 20 16 15 0 655 16 Format: BNEL rs, rt, offset MIPS32, removed in Release 6 Purpose: Branch on Not Equal Likely To compare GPRs then do a PC-relative conditional branch; execute the delay slot only if the branch is taken. Description: if GPR[rs]  GPR[rt] then branch_likely An 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), in the branch delay slot, to form a PC-relative effective target address. If the contents of GPR rs and GPR rt are not equal, branch to the effective target address after the instruction in the delay slot is executed. If the branch is not taken, the instruction in the delay slot is not executed. Restrictions: Processor operation is UNPREDICTABLE if a branch, jump, ERET, DERET, or WAIT instruction is placed in the delay slot of a branch or jump. Availability and Compatibility: This instruction has been removed in Release 6. Operation: I: target_offset  sign_extend(offset || 02) condition  (GPR[rs]  GPR[rt]) I+1: if condition then PC  PC + target_offset else NullifyCurrentInstruction() endif Exceptions: None Implementation Note: Some implementations always predict that the branch will be taken, and do not use nor do they update the branch internal processor branch prediction tables for this instruction. To maintain performance compatibility, future imple- mentations are encouraged to do the same. Programming Notes: With the 18-bit signed instruction offset, the conditional branch range is  128 KBytes. Use jump (J) or jump register (JR) to branch to addresses outside this range. In Pre-Release 6 implementations, software is strongly encouraged to avoid the use of the Branch Likely instructions, as they will be removed from a future revision of the MIPS Architecture. Some implementations always predict the branch will be taken, so there is a significant penalty if the branch is not taken. Software should only use this instruction when there is a very high probability (98% or more) that the branch will be taken. If the branch is not likely to be taken or if the probability of a taken branch is unknown, software is encouraged to use the BNE instruction instead. BNEL 010101 rs rt offset The MIPS32® Instruction Set Manual, Revision 6.06 104 BNEL Branch on Not Equal Likely Historical Information: In the MIPS I architecture, this instruction signaled a Reserved Instruction exception. 105 The MIPS32® Instruction Set Manual, Revision 6.06 BNEL IBranch on Not Equal Likely The MIPS32® Instruction Set Manual, Revision 6.06 106 BOVC BNVC Branch on Overflow, Compact; Branch on No Overflow, Compact 31 26 25 21 20 16 15 655 16 Format: BOVC BNVC BOVC rs,rt,offset BNVC rs,rt,offset Purpose: Branch on Overflow, Compact; Branch on No Overflow, Compact BOVC: Detect overflow for add (signed 32 bits) and branch if overflow. BNVC: Detect overflow for add (signed 32 bits) and branch if no overflow. 0 MIPS32 Release 6 MIPS32 Release 6 POP10 001000 BOVC rs >=rt
offset
rs
rt
POP30 011000
BNVC rs>=rt
offset
rs
rt
107
The MIPS32® Instruction Set Manual, Revision 6.06
Description:branchif/if-not NotWordValue(GPR[rs]+GPR[rt])
• BOVC performs a signed 32-bit addition of rs and rt. BOVC discards the sum, but detects signed 32-bit inte-
ger overflow of the sum, and branches if such overflow is detected.
• BNVC performs a signed 32-bit addition of rs and rt. BNVC discards the sum, but detects signed 32-bit inte- ger overflow of the sum, and branches if such overflow is not detected.
BOVC and BNVC are compact branches—they have no branch delay slots, but do have a forbidden slot.
A 18-bit signed offset (the 16-bit offset field shifted left 2 bits) is added to the address of the instruction following the branch (not the branch itself), to form a PC-relative effective target address.
The special case with rt=0 (for example, GPR[0]) is allowed. On MIPS32, BOVC rs,r0 offset never branches, while BNVC rs,r0 offset always branches.
The special case of rs=0 and rt=0 is allowed. BOVC never branches, while BNVC always branches.
Restrictions:
Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE.
If a control transfer instruction (CTI) is executed in the forbidden slot of a compact branch, Release 6 implementa- tions are required to signal a Reserved Instruction exception, but only when the branch is not taken.
Availability and Compatibility:
These instructions are introduced by and required as of Release 6.
See section A.4 on page 461 in Volume II for a complete overview of Release 6 instruction encodings. Brief notes related to these instructions:
• BOVC uses the primary opcode allocated to MIPS32 pre-Release 6 ADDI. Release 6 reuses the ADDI primary opcode for BOVC and other instructions, distinguished by register numbers.
• BNVC uses the primary opcode allocated to MIPS64 pre-Release 6 DADDI. Release 6 reuses the DADDI pri- mary opcode for BNVC and other instructions, distinguished by register numbers.
Operation:
temp1  GPR[rs] temp2  GPR[rt]

BOVC BNVC IBranch on Overflow, Compact; Branch on No Overflow, Compact tempd  temp1 + temp2 // wider than 32-bit precision
sum_overflow  (tempd32  tempd31) BOVC: cond  sum_overflow
BNVC: cond  not( sum_overflow )
if cond then
PC  ( PC+4 + sign_extend( offset << 2 ) ) endif Exceptions: None The MIPS32® Instruction Set Manual, Revision 6.06 108 BREAK Breakpoint 31 26 25 6 5 6 20 6 Format: BREAK Purpose: Breakpoint To cause a Breakpoint exception Description: 0 MIPS32 SPECIAL 000000 code BREAK 001101 109 The MIPS32® Instruction Set Manual, Revision 6.06 A breakpoint exception occurs, immediately and unconditionally transferring control to the exception handler. The code field is available for use as software parameters, but is retrieved by the exception handler only by loading the contents of the memory word containing the instruction. Restrictions: None Operation: SignalException(Breakpoint) Exceptions: Breakpoint C.cond.fmt IFloating Point Compare 31 26 25 21 20 16 15 11 10 8 7 6 5 4 3 0 655531124 COP1 010001 fmt ft fs cc 0 A 0 FC 11 cond Format: C.cond.fmt C.cond.S fs, ft (cc C.cond.D fs, ft (cc C.cond.PS fs, ft(cc C.cond.S cc, fs, ft C.cond.D cc, fs, ft C.cond.PS cc, fs, ft Purpose: Floating Point Compare To compare FP values and record the Boolean result in a condition code. = 0 implied) = 0 implied) = 0 implied) MIPS32, removed in Release 6 MIPS32, removed in Release 6 MIPS32 Release 2, removed in Release 6 MIPS32, removed in Release 6 MIPS32, removed in Release 6 MIPS32 Release 2, removed in Release 6 Description: FPConditionCode(cc)  FPR[fs] compare_cond FPR[ft] The value in FPR fs is compared to the value in FPR ft; the values are in format fmt. The comparison is exact and nei- ther overflows nor underflows. If the comparison specified by the cond field of the instruction is true for the operand values, the result is true; other- wise, the result is false. If no exception is taken, the result is written into condition code CC; true is 1 and false is 0. In the cond field of the instruction: cond2..1 specify the nature of the comparison (equals, less than, and so on). cond0 specifies whether the comparison is ordered or unordered, that is, false or true if any operand is a NaN; cond3 indi- cates whether the instruction should signal an exception on QNaN inputs, or not (see Table 3.2). C.cond.PS compares the upper and lower halves of FPR fs and FPR ft independently and writes the results into condi- tion codes CC +1 and CC respectively. The CC number must be even. If the number is not even the operation of the instruction is UNPREDICTABLE. If one of the values is an SNaN, or cond3 is set and at least one of the values is a QNaN, an Invalid Operation condi- tion is raised and the Invalid Operation flag is set in the FCSR. If the Invalid Operation Enable bit is set in the FCSR, no result is written and an Invalid Operation exception is taken immediately. Otherwise, the Boolean result is written into condition code CC. There are four mutually exclusive ordering relations for comparing floating point values; one relation is always true and the others are false. The familiar relations are greater than, less than, and equal. In addition, the IEEE floating point standard defines the relation unordered, which is true when at least one operand value is NaN; NaN compares unordered with everything, including itself. Comparisons ignore the sign of zero, so +0 equals -0. The comparison condition is a logical predicate, or equation, of the ordering relations such as less than or equal, equal, not less than, or unordered or equal. Compare distinguishes among the 16 comparison predicates. The Bool- ean result of the instruction is obtained by substituting the Boolean value of each ordering relation for the two FP val- ues in the equation. If the equal relation is true, for example, then all four example predicates above yield a true result. If the unordered relation is true then only the final predicate, unordered or equal, yields a true result. Logical negation of a compare result allows eight distinct comparisons to test for the 16 predicates as shown in Table 3.2. Each mnemonic tests for both a predicate and its logical negation. For each mnemonic, compare tests the truth of the first predicate. When the first predicate is true, the result is true as shown in the “If Predicate Is True” column, and the second predicate must be false, and vice versa. (Note that the False predicate is never true and False/True do not follow the normal pattern.) The truth of the second predicate is the logical negation of the instruction result. After a compare instruction, test for the truth of the first predicate can be made with the Branch on FP True (BC1T) instruction and the truth of the second The MIPS32® Instruction Set Manual, Revision 6.06 110 C.cond.fmt Floating Point Compare can be made with Branch on FP False (BC1F). Table 3.2 shows another set of eight compare operations, distinguished by a cond3 value of 1 and testing the same 16 conditions. For these additional comparisons, if at least one of the operands is a NaN, including Quiet NaN, then an Invalid Operation condition is raised. If the Invalid Operation condition is enabled in the FCSR, an Invalid Operation exception occurs. Table 3.1 FPU Comparisons Without Special Operand Exceptions Instruction Comparison Predicate Comparison CC Result Instruction Cond Mnemonic Name of Predicate and Logically Negated Predicate (Abbreviation) Relation Values If Predicate Is True Inv Op Excp. if QNaN? Condition Field >
< = ? 3 2..0 F False [this predicate is always False] F F F F F No 0 0 True (T) T T T T UN Unordered F F F T T 1 Ordered (OR) T T T F F EQ Equal F F T F T 2 Not Equal (NEQ) T T F T F UEQ Unordered or Equal F F T T T 3 Ordered or Greater Than or Less Than (OGL) T T F F F OLT Ordered or Less Than F T F F T 4 Unordered or Greater Than or Equal (UGE) T F T T F ULT Unordered or Less Than F T F T T 5 Ordered or Greater Than or Equal (OGE) T F T F F OLE Ordered or Less Than or Equal F T T F T 6 Unordered or Greater Than (UGT) T F F T F ULE Unordered or Less Than or Equal F T T T T 7 Ordered or Greater Than (OGT) T F F F F Key: ? = unordered, > = greater than, < = less than, = is equal, T = True, F = False 111 The MIPS32® Instruction Set Manual, Revision 6.06 C.cond.fmt IFloating Point Compare Table 3.2 FPU Comparisons With Special Operand Exceptions for QNaNs Instruction Comparison Predicate Comparison CC Result Instruction Cond Mnemonic Name of Predicate and Logically Negated Predicate (Abbreviation) Relation Values If Predicate Is True Inv Op Excp If QNaN? Condition Field >
< = ? 3 2..0 SF Signaling False [this predicate always False] F F F F F Yes 1 0 Signaling True (ST) T T T T NGLE Not Greater Than or Less Than or Equal F F F T T 1 Greater Than or Less Than or Equal (GLE) T T T F F SEQ Signaling Equal F F T F T 2 Signaling Not Equal (SNE) T T F T F NGL Not Greater Than or Less Than F F T T T 3 Greater Than or Less Than (GL) T T F F F LT Less Than F T F F T 4 Not Less Than (NLT) T F T T F NGE Not Greater Than or Equal F T F T T 5 Greater Than or Equal (GE) T F T F F LE Less Than or Equal F T T F T 6 Not Less Than or Equal (NLE) T F F T F NGT Not Greater Than F T T T T 7 Greater Than (GT) T F F F F Key: ? = unordered, > = greater than, < = less than, = is equal, T = True, F = False Restrictions: The fields fs and ft must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPRE- DICTABLE. The operands must be values in format fmt; if they are not, the result is UNPREDICTABLE and the value of the operand FPRs becomes UNPREDICTABLE. The result of C.cond.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model; it is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU,. The result of C.cond.PS is UNPREDICTABLE if the condition code number is odd. Availability and Compatibility: This instruction has been removed in Release 6 and has been replaced by the ‘CMP.cond fmt’ instruction. Refer to the CMP.cond fmt instruction in this manual for more information. Release 6 does not support Paired Single (PS). Operation: if SNaN(ValueFPR(fs, fmt)) or SNaN(ValueFPR(ft, fmt)) or QNaN(ValueFPR(fs, fmt)) or QNaN(ValueFPR(ft, fmt)) then less  false equal  false unordered  true if (SNaN(ValueFPR(fs,fmt)) or SNaN(ValueFPR(ft,fmt))) or (cond3 and (QNaN(ValueFPR(fs,fmt)) or QNaN(ValueFPR(ft,fmt)))) then The MIPS32® Instruction Set Manual, Revision 6.06 112 113 The MIPS32® Instruction Set Manual, Revision 6.06 C.cond.fmt Floating Point Compare SignalException(InvalidOperation) endif else less  ValueFPR(fs, fmt) <=? Predicates Long names IEEE Relation ><=? Negated Predicates Long names IEEE 0 0000 FFFF F FCAF AF False Always False TTTT T AT OR FCOR OR True Always True 1 0001 2 0010 FFFT FFTF UN FCUN EQ FCEQ UN EQ compareQuietUnordered Unordered ? TTTF TTFT Ordered compareQuietOrdered <=> NOT(isUnordered)
3 0011
FFTT
UEQ FCUEQ UEQ
Unordered or Equal
TTFF OGL
FCNE NE
Ordered Greater Than or Less Than
4 0100
FTFF
OLT FCLT
LT
Ordered Less Than compareQuietLess isLess
TFTT UGE
UGE
Unordered or Greater Than or Equal
compareQuietNotLess ?>=, NOT(isLess)
5 0101
FTFT
ULT FCULT
ULT
Unordered or Less compareQuietLessUnor-
TFTF OGE
OGE
Ordered Greater Than or Equal
compareQuiet- GreatrEqual isGreaterEqual
6 0110
7 0111
FTTF F T T T
OLE FCLE
ULE FCULE ULE
Ordered Less than or compareQuietLessEqual Equal isLessEqual
TFFT UGT T F F F OGT
UGT OGT
Unordered or Greater Than
compareQuietGreaterUn- ordered
Table 3.9 Comparing CMP.condn.fmt, IEEE 754-2008, C.cond.fmt, and MSA FP compares
Shaded entries in the table are unimplemented, and reserved.
LE
Instruction Encodings
CMP.condn.fmt: 010001 fffff ttttt sssss ddddd 0ccccc C.cond.fmt: 010001 fffff ttttt sssss CCC00 11cccc MSA: 011110 oooof ttttt sssss ddddd mmmmmm
Equal compareQuietEqual =
NEQ
FCUNE UNE
Not Equal
compareQuietNotEqual ?<>, NOT(=), 
Than dered
?<, NOT(isGreaterEqual) Unordered or Less compareQuietNotGreater Than or Equal ?<=, NOT(isGreater) Ordered Greater Than compareQuietGreater isGreater isUnordered ?>, NOT(isLessEqual)
no (non-signalling) yes (always signal SNaN)
Invalid Operand Exception
C condn.fmt
MSA
CMP
condn.fmt
C condn.fmt
MSA
CMP
condn.fmt

CMP.condn.fmt Floating Point Compare Setting Mask
139
The MIPS32® Instruction Set Manual, Revision 6.06
.
MSA: operation
MSA: minor opcode mmmmmm Bits 5…0 = 26 – 011010 CMP: condn Bit 5..4 = 00 C: only applicable
MSA: minor opcode mmmmmm Bits 5…0 = 28 – 011100 CMP: condn Bit 5..4 = 01 C: not applicable
oooo Bits 25…22 C: cond cccc – Bits 3..0 CMP: condn cccccc – Bits 3..0
Relation ><=? Predicates Long names IEEE Relation ><=? Negated Predicates Long names IEEE 11 1011 F F T T NGL FSUEQ SUEQ Signalling Unordered or Equal T T F F GL FSNE SNE Signalling Ordered 12 1100 F T F F LT FSLT SLT T F T T Unordered or Greater Than or Equal 13 1101 F T F T NGE FSULT SULT Not Greater Than or Equal Unordered or Less compareSignalling- LessUnordered T F T F GE SOGE Signalling Ordered Greater Than or Equal compareSignalling- GreaterEqual 14 1110 F T T F LE FSLE SLE Ordered Signalling LessEqual T F F T NLE SUGT Not Less Than or Equal GreaterUnordered 15 1111 F T T T NGT FSULE SULE Not Greater Than Signalling Unordered NotGreater NOT(>)
T
F
F
F
GT SOGT
Greater Than Signalling Ordered
compareSignalling- Greater
Table 3.9 Comparing CMP.condn.fmt, IEEE 754-2008, C.cond.fmt, and MSA FP compares (Continued)
Shaded entries in the table are unimplemented, and reserved.
8 1000 FFFF
SF FSAF SAF
Signalling False Signalling
TTTT
ST
SAT
Signalling True Signalling
9 1001 F F
F T
NGLE FSUN SUN
Not Greater Than or Less Than or Equal
T
T
T
F
GLE FSOR SOR
Greater Than or
Less Than or Equa
10 1010 F F
T F
SEQ FSEQ SEQ
compareSignalling Equal
T
T
F
T
SNE FSUNE SUNE
compareSignalling- dered or Not NotEqual
Always False
Always True
Signalling Unordered
Signalling Ordered
Signalling Equal Ordered Signalling
Signalling Not Equal Signalling Unor-
Equal
Equal
Not Greater Than or Less Than
Greater Than or Less Than
Less Than
Ordered Signalling
compareSignallingLess < NLT SUGE Not Less Than Signalling compareSignallingNot- Less Less Than NOT(<) Than Less Than or Equal NOT(>=) compareSignalling-
>=,  compareSignalling-
Less Than or Equal
<=,  compareSignalling- Signalling Unordered or Greater Than NOT(<=) or Less Than or Equal Greater Than >
Instruction Encodings
CMP.condn.fmt: 010001 fffff ttttt sssss ddddd 0ccccc C.cond.fmt: 010001 fffff ttttt sssss CCC00 11cccc MSA: 011110 oooof ttttt sssss ddddd mmmmmm
Not Equal
yes (signalling)
Invalid Operand Exception
C condn.fmt
MSA
CMP
condn.fmt
C condn.fmt
MSA
CMP
condn.fmt

140
The MIPS32® Instruction Set Manual, Revision 6.06
CMP.condn.fmt Floating Point Compare Setting Mask Restrictions:
Operation:
if SNaN(ValueFPR(fs, fmt)) or SNaN(ValueFPR(ft, fmt)) or
QNaN(ValueFPR(fs, fmt)) or QNaN(ValueFPR(ft, fmt))
then
less  false
equal  false
unordered  true
if (SNaN(ValueFPR(fs,fmt)) or SNaN(ValueFPR(ft,fmt))) or
(cond3 and (QNaN(ValueFPR(fs,fmt)) or QNaN(ValueFPR(ft,fmt)))) then SignalException(InvalidOperation)
endif else
less  ValueFPR(fs, fmt) > 1) xor poly
else
value  (value >> 1)
endif endfor
return value
endfunction
Exceptions:
Reserved Instruction Exception
Restriction:
These instructions are implemented in Release 6 only if CP0 Config5CRCP is set to 1. Programing Notes:
When calculating CRC, it is recommended that the initial value of GPR[rt] be all ones, when the CRC instruction is the first in the sequence to be referenced. This allows the CRC to differentiate between actual leading zeroes in the
The MIPS32® Instruction Set Manual, Revision 6.06 144

CRC32B, CRC32H, CRC32W IGenerate CRC with reversed polynomial 0xEDB88320
message element, and zeros added by transmission errors. The initial all one’s value makes no difference to the CRC calculation as long as both sender and receiver use the same assumption on the initial value, to generate and check respectively.
If the order of bits in bytes assumes most-significant bit first, then Release 6 BITSWAP can be used to reverse the order of bits in order to operate with these instructions. However BITSWAP would only apply to byte messages.
CRC32B/H/Winstructions are interchangeable: a series of low-order CRC instructions can be reduced to a series of high-order CRC32H operations, to increase throughput of the overall CRC generation process. The process of doing this will add trailing zeroes to the message for which CRC is being generated, since the data element is now larger, however, this will not change the CRC value that is generated. It is the original message that must be transmitted along with the CRC, without the trailing zeroes.
In pseudo-assembly, the following sequence of byte CRC operations may be used to generate a cumulative CRC value. (Pseudo-assembly is used to clearly indicate terms which need to be modified for interchangeability.)
li $3, 0xFFFF_FFFF // initialize CRC value
la $4, memaddr // assume word-aligned for convenience
for (i=0; i < byte_cnt; i++) lb $2, 0($4) // read message bytes crc32b $3, $2, $3 add $4, $4, 1 // increment byte memory address by 1 endfor This is equivalent to the sequence of word CRC operations. The simple example assumes some multiple of 4 bytes are processed. for (i=0; i < byte_cnt/4; i++) lw $2, 0($4) // read message words crc32w $3, $2, $3 add $4, $4, 4 // increment word memory address by 4 endfor The throughput is thus increased by a multiple of 4 as only a quarter of the byte oriented operations occur. The MIPS32® Instruction Set Manual, Revision 6.06 145 CRC32CB, CRC32CH, CRC32CW IGenerate CRC with reversed polynomial 0x82F63B78 31 26 25 21 20 16 15 11 10 6 5 0 6555326 SPECIAL3 011111 rs rt 00000 001 sz CRC 001111 Format: CRC32CB, CRC32CH, CRC32CW CRC32CB rt, rs, rt CRC32CH rt, rs, rt CRC32CW rt, rs, rt Purpose: Generate CRC with reversed polynomial 0x82F63B78 Description: GPR[rt]  CRC32C(GRP[rs], GPR[rt]) MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 CRC32CB/H/W generates a 32-bit Cyclic Redundancy Check (CRC) value based on the reversed polynomial 0x82F63B78 (Castagnoli). The new 32-bit CRC value is generated with a cumulative 32-bit CRC value input as GPR[rt] and a byte or half-word or word message right-justified in GPR[rs]. The message size is encoded in field sz of the instruction. The generated value overwrites the input CRC value in GPR[rt], as the original value is considered redundant once the cumulative CRC value is re-generated with the additional message. More importantly, source-destroying defini- tion of the CRC instruction allows the instruction to be included in a loop without having to move the destination to the source for the next iteration of the instruction. The CRC32CB/H/W instruction does not pad the input message. It is software’s responsibility to ensure the input message, whether byte, half-word, word or double-word is fully-defined, otherwise the result is UNPREDICTABLE and thus unusable. The reversed polynomial is a 33-bit polynomial of degree 32. Since the coefficient of most significance is always 1, it is dropped from the 32-bit binary number, as per standard representation. The order of the remaining coefficients increases from right to left in the binary representation. Since the CRC is processed more than a bit at a time, the order of bits in the data elements of size byte, half-word, word or double-word is important. The specification assumes support for an “lsb-first” (little-endian) standard, and thus coefficients of polynomial terms that represent the message must be ordered from right to left in order of dec- creasing significance. The specification of the CRC instruction assumes the following in regards to a message of arbitrary length whose 32- bit CRC value is to be generated. The message itself is a polynomial represented by binary coefficients of each term of the polynomial. • The message is a sequence of bytes or half-words or words as per use case. The appropriate instruction is chosen. • For each message element of size byte/half-word/word/double-word, the least-significant bit corresponds to the most significant coefficient, and significance decreases from right to left. • Message elements themselves must be processed in order of decreasing significance, with reference to coeffi- cients of the terms of the polynomial the message represents. • The polynomial is thus reversed to match the order of coefficients for the message of arbitrary length. • The resultant CRC is also a polynomial whose coefficients are arranged in decreasing significance from right to left. The MIPS32® Instruction Set Manual, Revision 6.06 146 CRC32CB, CRC32CH, CRC32CW IGenerate CRC with reversed polynomial 0x82F63B78 The typical use of CRC is to generate a checksum to accompany a message that is transmitted electronically in order to detect transmission errors at the receiving end. If the message is considered to be a polynomial, the coefficient of the most-significant term is transmitted first followed by remaining bits in order of decreasing significance, followed by the 32-bit CRC. The specification for these CRC instruction is thus most appropriate for standards that transmit least significant bit of data first (little-endian), such as IEEE 802 Ethernet. The least-significant bit of data conve- niently maps to the coefficient of the most-significant term of the message polynomial. Restrictions: No data-dependent exceptions are possible. Operation: if (Config5CRCP = 0) then SignalException(ReservedInstruction) endif if (sz = 0b00) then temp  CRC32(GPR[rt], GPR[rs], 1, 0x82F63B78) else if (sz = 0b01) then temp  CRC32(GPR[rt], GPR[rs], 2, 0x82F63B78) else if (sz = 0b10) then temp  CRC32(GPR[rt], GPR[rs], 4, 0x82F63B78) else if (sz = 0b11) then SignalException(ReservedInstruction) endif GPR[rt] temp // Bit oriented definition of CRC32 function function CRC32(value, message, numbytes, poly) # value - right-justified current 32-bit CRC value # message - right-justified byte/half-word/word message # numbytes - size of message in bytes: byte/half-word/word # poly - 32-bit reversed polynomial value  value xor {(32-(numbytes*8))’b0,message} for (i=0; i> 1) xor poly
else
value  (value >> 1)
endif endfor
return value
endfunction
Exceptions:
Reserved Instruction Exception
Restriction:
These instructions are implemented in Release 6 only if CP0 Config5CRCP is set to 1. Programing Notes:
When calculating CRC, it is recommended that the initial value of GPR[rt] be all ones, when the CRC instruction is the first in the sequence to be referenced. This allows the CRC to differentiate between actual leading zeroes in the
The MIPS32® Instruction Set Manual, Revision 6.06 147

CRC32CB, CRC32CH, CRC32CW IGenerate CRC with reversed polynomial 0x82F63B78
message element, and zeros added by transmission errors. The initial all one’s value makes no difference to the CRC calculation as long as both sender and receiver use the same assumption on the initial value, to generate and check respectively.
If the order of bits in bytes assumes most-significant bit first, then Release 6 BITSWAP can be used to reverse the order of bits in order to operate with these instructions. However BITSWAP would only apply to byte messages.
CRC32CB/H/Winstructions are interchangeable: a series of low-order CRC instructions can be reduced to a series of high-order CRC32CH operations, to increase throughput of the overall CRC generation process. The process of doing this will add trailing zeroes to the message for which CRC is being generated, since the data element is now larger, however, this will not change the CRC value that is generated. It is the original message that must be transmitted along with the CRC, without the trailing zeroes.
In pseudo-assembly, the following sequence of byte CRC operations may be used to generate a cumulative CRC value. (Pseudo-assembly is used to clearly indicate terms which need to be modified for interchangeability.)
li $3, 0xFFFF_FFFF // initialize CRC value
la $4, memaddr // assume word-aligned for convenience
for (i=0; i < byte_cnt; i++) lb $2, 0($4) // read message bytes crc32cb $3, $2, $3 add $4, $4, 1 // increment byte memory address by 1 endfor This is equivalent to the sequence of word CRC operations. The simple example assumes some multiple of 4 bytes are processed. for (i=0; i < byte_cnt/4; i++) lw $2, 0($4) // read message words crc32cw $3, $2, $3 add $4, $4, 4 // increment word memory address by 4 endfor The throughput is thus increased by a multiple of 4 as only a quarter of the byte oriented operations occur. The MIPS32® Instruction Set Manual, Revision 6.06 148 CTC1 IMove Control Word to Floating Point 31 26 25 21 20 16 15 11 10 0 6 5 5 5 11 Format: CTC1 rt, fs MIPS32 Purpose: Move Control Word to Floating Point To copy a word from a GPR to an FPU control register. Description: FP_Control[fs]  GPR[rt] Copy the low word from GPR rt into the FP (coprocessor 1) control register indicated by fs. Writing to the floating point Control/Status register, the FCSR, causes the appropriate exception if any Cause bit and its corresponding Enable bit are both set. The register is written before the exception occurs. Writing to FEXR to set a cause bit whose enable bit is already set, or writing to FENR to set an enable bit whose cause bit is already set causes the appropriate exception. The register is written before the exception occurs and the EPC register contains the address of the CTC1 instruction. The definition of this instruction has been extended in Release 5 to support user mode read and write of StatusFR under the control of Config5UFR. This optional feature is meant to facilitate transition from FR=0 to FR=1 float- ing-point register modes in order to obsolete FR=0 mode in a future architecture release. User code may set and clear StatusFR without kernel intervention, providing kernel explicitly provides permission. This UFR facility is not supported in Release 6 since Release 6 only allows FR=1 mode. Accessing the UFR and UNFR registers causes a Reserved Instruction exception in Release 6 since FIRUFRP is always 0. The definition of this instruction has been extended in Release 6 to allow user code to read and modify the Config5FRE bit. Such modification is allowed when this bit is present (as indicated by FIRUFRP) and user mode modification of the bit is enabled by the kernel (as indicated by Config5UFE). Setting Config5FRE to 1 causes all floating point instructions which are not compatible with FR=1 mode to take an Reserved Instruction exception. This makes it possible to run pre-Release 6 FR=0 floating point code on a Release 6 core which only supports FR=1 mode, provided the kernel has been set up to trap and emulate FR=0 behavior for these instructions. These instructions include floating-point arithmetic instructions that read/write single-precision registers, LWC1, SWC1, MTC1, and MFC1 instructions. The FRE facility uses COP1 register aliases FRE and NFRE to access Config5FRE. Restrictions: There are a few control registers defined for the floating point unit. Prior to Release 6, the result is UNPREDICT- ABLE if fs specifies a register that does not exist. In Release 6 and later, a Reserved Instruction exception occurs if fs specifies a register that does not exist. Furthermore, the result is UNPREDICTABLE if fd specifies the UFR, UNFR, FRE and NFRE aliases, with fs any- thing other than 00000, GPR[0]. Release 6 implementations and later are required to produce a Reserved Instruction exception; software must assume it is UNPREDICTABLE. Operation: temp  GPR[rt]31..0 if (fs = 1 or fs = 4) then /* clear UFR or UNFR(CP1 Register 1)*/ if ConfigAR ≥ 2 SignalException(ReservedInstruction) /* Release 6 traps */ endif COP1 010001 CT 00110 rt fs 0 000 0000 0000 The MIPS32® Instruction Set Manual, Revision 6.06 149 150 The MIPS32® Instruction Set Manual, Revision 6.06 CTC1 Move Control Word to Floating Point if not Config5UFR then SignalException(ReservedInstruction) endif if not (rt = 0 and FIRUFRP) then UNPREDICTABLE /*end of instruction*/ endif if fs = 1 then StatusFR  0 elseif fs = 4 then StatusFR  1 else /* cannot happen */ elseif fs=5 then /* user write of 1 to FRE, if permitted */ if ConfigAR  2 then UNPREDICTABLE else if rt ≠ 0 then SignalException(ReservedInstruction) endif if not Config5UFR then SignalException(ReservedInstruction) endif Config5UFR  0 endif elseif fs=6 then /* user write of 0 to FRE, if permitted (NFRE alias) */ if ConfigAR  2 then UNPREDICTABLE else if rt ≠ 0 then SignalException(ReservedInstruction) endif if not Config5UFR then SignalException(ReservedInstruction) endif Config5UFR  1 endif elseif fs = 25 then /* FCCR */ if temp31..8 ≠ 024 then UNPREDICTABLE else FCSR  temp7..1 || FCSR24 || temp0 || FCSR22..0 endif elseif fs = 26 then /* FEXR */ if temp31..18 ≠ 0 or temp11..7 ≠ 0 or temp2..0 ≠ 0then UNPREDICTABLE else FCSR  FCSR31..18 || temp17..12 || FCSR11..7 || temp6..2 || FCSR1..0 endif elseif fs = 28 then /* FENR */ if temp31..12 ≠ 0 or temp6..3 ≠ 0 then UNPREDICTABLE else FCSR  FCSR31..25 || temp2 || FCSR23..12 || temp11..7 || FCSR6..2 || temp1..0 endif elseif fs = 31 then /* FCSR */ if (FCSRImpl field is not implemented) and(temp22..18 ≠ 0) then UNPREDICTABLE elseif (FCSRImpl field is implemented) and temp20..18 ≠ 0 then UNPREDICTABLE else FCSR  temp endif else if Config2AR ≥ 2 SignalException(ReservedInstruction) /* Release 6 traps */ endif UNPREDICTABLE endif CheckFPException() Exceptions: Coprocessor Unusable, Reserved Instruction CTC1 IMove Control Word to Floating Point Unimplemented Operation, Invalid Operation, Division-by-zero, Inexact, Overflow, Underflow Historical Information: For the MIPS I, II and III architectures, the contents of floating point control register fs are UNPREDICTABLE for the instruction immediately following CTC1. MIPS V and MIPS32 introduced the three control registers that access portions of FCSR. These registers were not available in MIPS I, II, III, or IV. MIPS32 Release 5 introduced the UFR and UNFR register aliases that allow user level access to StatusFR. MIPS32 Release 6 introduced the FRE and NFRE register aliases that allow user to cause traps for FR=0 mode emu- lation. Floating Point Exceptions: The MIPS32® Instruction Set Manual, Revision 6.06 151 CTC2 Move Control Word to Coprocessor 2 31 26 25 21 20 16 15 11 10 0 655 16 Format: CTC2 rt, Impl MIPS32 The syntax shown above is an example using CTC1 as a model. The specific syntax is implementation dependent. Purpose: Move Control Word to Coprocessor 2 To copy a word from a GPR to a Coprocessor 2 control register. Description: CP2CCR[Impl]  GPR[rt] Copy the low word from GPR rt into the Coprocessor 2 control register denoted by the Impl field. The interpretation of the Impl field is left entirely to the Coprocessor 2 implementation and is not specified by the architecture. Restrictions: The result is UNPREDICTABLE if rd specifies a register that does not exist. Operation: temp  GPR[rt] CP2CCR[Impl]  temp Exceptions: Coprocessor Unusable, Reserved Instruction COP2 010010 CT 00110 rt Impl 152 The MIPS32® Instruction Set Manual, Revision 6.06 CVT.D.fmt IFloating Point Convert to Double Floating Point 31 26 25 21 20 16 15 11 10 6 5 655556 Format: CVT.D.fmt CVT.D.S fd, fs CVT.D.W fd, fs CVT.D.L fd, fs Purpose: Floating Point Convert to Double Floating Point To convert an FP or fixed point value to double FP. Description: FPR[fd]  convert_and_round(FPR[fs]) 0 MIPS32 MIPS32 MIPS32 Release 2 COP1 010001 fmt 0 00000 fs fd CVT.D 100001 The value in FPR fs, in format fmt, is converted to a value in double floating point format and rounded according to the current rounding mode in FCSR. The result is placed in FPR fd. If fmt is S or W, then the operation is always exact. Restrictions: The fields fs and fd must specify valid FPRs, fs for type fmt and fd for double floating point. If the fields are not valid, the result is UNPREDICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. For CVT.D.L, the result of this instruction is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. Operation: StoreFPR (fd, D, ConvertFmt(ValueFPR(fs, fmt), fmt, D)) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Invalid Operation, Unimplemented Operation, Inexact The MIPS32® Instruction Set Manual, Revision 6.06 153 CVT.L.fmt Floating Point Convert to Long Fixed Point 31 26 25 21 20 16 15 11 10 6 5 0 655556 COP1 010001 fmt 0 00000 fs fd CVT.L 100101 154 The MIPS32® Instruction Set Manual, Revision 6.06 Format: CVT.L.fmt CVT.L.S fd, fs CVT.L.D fd, fs Purpose: Floating Point Convert to Long Fixed Point To convert an FP value to a 64-bit fixed point. MIPS32 Release 2 MIPS32 Release 2 Description: FPR[fd]  convert_and_round(FPR[fs]) Convert the value in format fmt in FPR fs to long fixed point format and round according to the current rounding mode in FCSR. The result is placed in FPR fd. When the source value is Infinity, NaN, or rounds to an integer outside the range -263 to 263-1, the result cannot be represented correctly, an IEEE Invalid Operation condition exists, and the Invalid Operation flag is set in the FCSR. If the Invalid Operation Enable bit is set in the FCSR, no result is written to fd and an Invalid Operation exception is taken immediately. Otherwise, a default result is written to fd. On cores with FCSRNAN2008=0, the default result is 263–1. On cores with FCSRNAN2008=1, the default result is: • 0 when the input value is NaN • 263–1 when the input value is + or rounds to a number larger than 263–1 • -263–1 when the input value is – or rounds to a number smaller than -263–1 Restrictions: The fields fs and fd must specify valid FPRs, fs for type fmt and fd for long fixed point. If the fields are not valid, the result is UNPREDICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. The result of this instruction is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model; it is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Operation: StoreFPR (fd, L, ConvertFmt(ValueFPR(fs, fmt), fmt, L)) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Invalid Operation, Unimplemented Operation, Inexact, CVT.PS.S Floating Point Convert Pair to Paired Single 156 The MIPS32® Instruction Set Manual, Revision 6.06 CVT.S.PL IFloating Point Convert Pair Lower to Single Floating Point 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: CVT.S.PL fd, fs MIPS32 Release 2, removed in Release 6 Purpose: Floating Point Convert Pair Lower to Single Floating Point To convert one half of a paired single FP value to single FP. Description: FPR[fd]  FPR[fs]31..0 The lower paired single value in FPR fs, in format PS, is converted to a value in single floating point format. The result is placed in FPR fd. This instruction can be used to isolate the lower half of a paired single value. The operation is non-arithmetic; it causes no IEEE 754 exceptions, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs and fd must specify valid FPRs—fs for type PS and fd for single floating point. If the fields are not valid, the result is UNPREDICTABLE. The operand must be a value in format PS; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. The result of CVT.S.PL is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model; it is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: This instruction has been removed in Release 6. Operation: StoreFPR (fd, S, ConvertFmt(ValueFPR(fs, PS), PL, S)) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: COP1 010001 fmt 10110 0 00000 fs fd CVT.S.PL 101000 The MIPS32® Instruction Set Manual, Revision 6.06 157 CVT.S.PU Floating Point Convert Pair Upper to Single Floating Point 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: CVT.S.PU fd, fs MIPS32 Release 2, , removed in Release 6 Purpose: Floating Point Convert Pair Upper to Single Floating Point To convert one half of a paired single FP value to single FP Description: FPR[fd]  FPR[fs]63..32 The upper paired single value in FPR fs, in format PS, is converted to a value in single floating point format. The result is placed in FPR fd. This instruction can be used to isolate the upper half of a paired single value. The operation is non-arithmetic; it causes no IEEE 754 exceptions, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs and fd must specify valid FPRs—fs for type PS and fd for single floating point. If the fields are not valid, the result is UNPREDICTABLE. The operand must be a value in format PS; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. The result of CVT.S.PU is UNPREDICTABLE if the processor is executing the FR=0 32-bit FPU register model; it is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU Availability and Compatibility: This instruction was removed in Release 6. Operation: StoreFPR (fd, S, ConvertFmt(ValueFPR(fs, PS), PU, S)) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: COP1 010001 fmt 10110 0 00000 fs fd CVT.S.PU 100000 158 The MIPS32® Instruction Set Manual, Revision 6.06 CVT.S.fmt IFloating Point Convert to Single Floating Point 31 26 25 21 20 16 15 11 10 6 5 655556 Format: CVT.S.fmt CVT.S.D fd, fs CVT.S.W fd, fs CVT.S.L fd, fs Purpose: Floating Point Convert to Single Floating Point To convert an FP or fixed point value to single FP. 0 MIPS32 MIPS32 MIPS32 Release 2 COP1 010001 fmt 0 00000 fs fd CVT.S 100000 Description: FPR[fd]  convert_and_round(FPR[fs]) The value in FPR fs, in format fmt, is converted to a value in single floating point format and rounded according to the current rounding mode in FCSR. The result is placed in FPR fd. Restrictions: The fields fs and fd must specify valid FPRs—fs for type fmt and fd for single floating point. If the fields are not valid, the result is UNPREDICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. For CVT.S.L, the result of this instruction is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model; it is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Operation: StoreFPR(fd, S, ConvertFmt(ValueFPR(fs, fmt), fmt, S)) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Invalid Operation, Unimplemented Operation, Inexact, Overflow, Underflow The MIPS32® Instruction Set Manual, Revision 6.06 159 CVT.W.fmt Floating Point Convert to Word Fixed Point 31 26 25 21 20 16 15 11 10 6 5 655556 Format: CVT.W.fmt CVT.W.S fd, fs CVT.W.D fd, fs Purpose: Floating Point Convert to Word Fixed Point To convert an FP value to 32-bit fixed point. 0 MIPS32 MIPS32 COP1 010001 fmt 0 00000 fs fd CVT.W 100100 160 The MIPS32® Instruction Set Manual, Revision 6.06 Description: FPR[fd]  convert_and_round(FPR[fs]) The value in FPR fs, in format fmt, is converted to a value in 32-bit word fixed point format and rounded according to the current rounding mode in FCSR. The result is placed in FPR fd. When the source value is Infinity, NaN, or rounds to an integer outside the range -231 to 231-1, the result cannot be represented correctly, an IEEE Invalid Operation condition exists, and the Invalid Operation flag is set in the FCSR. If the Invalid Operation Enable bit is set in the FCSR, no result is written to fd and an Invalid Operation exception is taken immediately. Otherwise, a default result is written to fd. On cores with FCSRNAN2008=0, the default result is 263–1. On cores with FCSRNAN2008=1, the default result is: • 0 when the input value is NaN • 263–1 when the input value is + or rounds to a number larger than 263–1 • -263–1 when the input value is – or rounds to a number smaller than -263–1 Restrictions: The fields fs and fd must specify valid FPRs: fs for type fmt and fd for word fixed point. If the fields are not valid, the result is UNPREDICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. Operation: StoreFPR(fd, W, ConvertFmt(ValueFPR(fs, fmt), fmt, W)) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Invalid Operation, Unimplemented Operation, Inexact DDIV IDoubleword Divide 31 26 25 21 20 16 15 0 6 5 5 10 6 Format: DDIV rs, rt MIPS64, removed in Release 6 Purpose: Doubleword Divide To divide 64-bit signed integers. Description: (LO, HI)  GPR[rs] / GPR[rt] The 64-bit doubleword in GPR rs is divided by the 64-bit doubleword in GPR rt, treating both operands as signed val- ues. The 64-bit quotient is placed into special register LO and the 64-bit remainder is placed into special register HI. No arithmetic exception occurs under any circumstances. Restrictions: If the divisor in GPR rt is zero, the arithmetic result value is UNPREDICTABLE. Availability and Compatibility: This instruction has been removed in Release 6. Operation: LO  GPR[rs] div GPR[rt] HI  GPR[rs] mod GPR[rt] Exceptions: Reserved Instruction Programming Notes: See “Programming Notes” for the DIV instruction. Historical Perspective: In MIPS III, if either of the two instructions preceding the divide is an MFHI or MFLO, the result of the MFHI or MFLO is UNPREDICTABLE. Reads of the HI or LO special register must be separated from subsequent instruc- tions that write to them by two or more instructions. This restriction was removed in MIPS IV and MIPS32 and all subsequent levels of the architecture. SPECIAL 000000 rs rt 0 00 0000 0000 DDIV 011110 The MIPS32® Instruction Set Manual, Revision 6.06 161 DDIVU IDoubleword Divide Unsigned 31 26 25 21 20 16 15 6 5 0 6 5 5 10 6 Format: DDIVU rs, rt MIPS64, removed in Release 6 Purpose: Doubleword Divide Unsigned To divide 64-bit unsigned integers. Description: (LO, HI)  GPR[rs] / GPR[rt] The 64-bit doubleword in GPR rs is divided by the 64-bit doubleword in GPR rt, treating both operands as unsigned values. The 64-bit quotient is placed into special register LO and the 64-bit remainder is placed into special register HI. No arithmetic exception occurs under any circumstances. Restrictions: If the divisor in GPR rt is zero, the arithmetic result value is UNPREDICTABLE. Availability and Compatibility: This instruction has been removed in Release 6. Operation: q  (0 || GPR[rs]) div (0 || GPR[rt]) r  (0 || GPR[rs]) mod (0 || GPR[rt]) LO  q63..0 HI  r63..0 Exceptions: Reserved Instruction Programming Notes: See “Programming Notes” for the DIV instruction. Historical Perspective: In MIPS III, if either of the two instructions preceding the divide is an MFHI or MFLO, the result of the MFHI or MFLO is UNPREDICTABLE. Reads of the HI or LO special register must be separated from subsequent instruc- tions that write to them by two or more instructions. This restriction was removed in MIPS IV and MIPS32 and all subsequent levels of the architecture. SPECIAL 000000 rs rt 0 00 0000 0000 DDIVU 011111 The MIPS32® Instruction Set Manual, Revision 6.06 162 DERET IDebug Exception Return 31 26 25 24 6 5 0 61 19 6 Format: DERET EJTAG Purpose: Debug Exception Return To Return from a debug exception. Description: DERET clears execution and instruction hazards, returns from Debug Mode and resumes non-debug execution at the instruction whose address is contained in the DEPC register. DERET does not execute the next instruction (i.e. it has no delay slot). Restrictions: A DERET placed between an LL and SC instruction does not cause the SC to fail. If the DEPC register with the return address for the DERET was modified by an MTC0 or a DMTC0 instruction, a CP0 hazard exists that must be removed via software insertion of the appropriate number of SSNOP instructions (for implementations of Release 1 of the Architecture) or by an EHB, or other execution hazard clearing instruction (for implementations of Release 2 of the Architecture). DERET implements a software barrier that resolves all execution and instruction hazards created by Coprocessor 0 state changes (for Release 2 implementations, refer to the SYNCI instruction for additional information on resolving instruction hazards created by writing the instruction stream). The effects of this barrier are seen starting with the instruction fetch and decode of the instruction at the PC to which the DERET returns. This instruction is legal only if the processor is executing in Debug Mode. Pre-Release 6: The operation of the processor is UNDEFINED if a DERET is executed in the delay slot of a branch or jump instruction. Release 6 implementations are required to signal a Reserved Instruction exception if DERET is encountered in the delay slot or forbidden slot of a branch or jump instruction. Operation: DebugDM  0 DebugIEXI  0 if IsMIPS16Implemented() | (Config3ISA > 0) then
PC  DEPC31..1 || 0
ISAMode  DEPC0 else
PC  DEPC endif
ClearHazards()
Exceptions:
Coprocessor Unusable, Reserved Instruction
COP0 010000
CO 1
0
000 0000 0000 0000 0000
DERET 011111
The MIPS32® Instruction Set Manual, Revision 6.06 163

DI
Disable Interrupts
31
26 25 21 20 16 15 11 10 6 5 4 3 2 0
65555123
Format: DI MIPS32Release2 DI rt MIPS32 Release 2
Purpose: Disable Interrupts
To return the previous value of the Status register and disable interrupts. If DI is specified without an argument, GPR
r0 is implied, which discards the previous value of the Status register. Description: GPR[rt]  Status; StatusIE  0
The current value of the Status register is loaded into general register rt. The Interrupt Enable (IE) bit in the Status register is then cleared.
Restrictions:
If access to Coprocessor 0 is not enabled, a Coprocessor Unusable Exception is signaled.
In implementations prior to Release 2 of the architecture, this instruction resulted in a Reserved Instruction exception.
Operation:
This operation specification is for the general interrupt enable/disable operation, with the sc field as a variable. The individual instructions DI and EI have a specific value for the sc field.
data  Status GPR[rt]  data StatusIE  0
Exceptions:
Coprocessor Unusable
Reserved Instruction (Release 1 implementations)
Programming Notes:
The effects of this instruction are identical to those accomplished by the sequence of reading Status into a GPR, clearing the IE bit, and writing the result back to Status. Unlike the multiple instruction sequence, however, the DI instruction cannot be aborted in the middle by an interrupt or exception.
This instruction creates an execution hazard between the change to the Status register and the point where the change to the interrupt enable takes effect. This hazard is cleared by the EHB, JALR.HB, JR.HB, or ERET instructions. Soft- ware must not assume that a fixed latency will clear the execution hazard.
COP0 0100 00
MFMC0 01 011
rt
12 0110 0
0 000 00
sc 0
0 00
0 000
164
The MIPS32® Instruction Set Manual, Revision 6.06

DIV
IDivide Word
31
26 25 21 20 16 15 6 5 0
6 5 5 10 6
Format: DIV rs, rt MIPS32, removed in Release 6 Purpose: Divide Word
To divide a 32-bit signed integers.
Description: (HI, LO)  GPR[rs] / GPR[rt]
The 32-bit word value in GPR rs is divided by the 32-bit value in GPR rt, treating both operands as signed values.
The 32-bit quotient is placed into special register LO and the 32-bit remainder isplaced into special register HI. No arithmetic exception occurs under any circumstances.
Restrictions:
If the divisor in GPR rt is zero, the arithmetic result value is UNPREDICTABLE.
Availability and Compatibility:
DIV has been removed in Release 6 and has been replaced by DIV and MOD instructions that produce only quotient and remainder, respectively. Refer to the Release 6 introduced ‘DIV’ and ‘MOD’ instructions in this manual for more information. This instruction remains current for all release levels lower than Release 6 of the MIPS architecture.
Operation:
q  GPR[rs]31..0 div GPR[rt]31..0 LO  q
r  GPR[rs]31..0 mod GPR[rt]31..0 HI  r
Exceptions:
None
Programming Notes:
No arithmetic exception occurs under any circumstances. If divide-by-zero or overflow conditions are detected and some action taken, then the divide instruction is followed by additional instructions to check for a zero divisor and/or for overflow. If the divide is asynchronous then the zero-divisor check can execute in parallel with the divide. The action taken on either divide-by-zero or overflow is either a convention within the program itself, or within the sys- tem software. A possibility is to take a BREAK exception with a code field value to signal the problem to the system software.
As an example, the C programming language in a UNIX® environment expects division by zero to either terminate the program or execute a program-specified signal handler. C does not expect overflow to cause any exceptional con- dition. If the C compiler uses a divide instruction, it also emits code to test for a zero divisor and execute a BREAK instruction to inform the operating system if a zero is detected.
By default, most compilers for the MIPS architecture emits additional instructions to check for the divide-by-zero and overflow cases when this instruction is used. In many compilers, the assembler mnemonic “DIV r0, rs, rt” can be used to prevent these additional test instructions to be emitted.
In some processors the integer divide operation may proceed asynchronously and allow other CPU instructions to execute before it is complete. An attempt to read LO or HI before the results are written interlocks until the results are
SPECIAL 000000
rs
rt
0
00 0000 0000
DIV 011010
The MIPS32® Instruction Set Manual, Revision 6.06 165

166
The MIPS32® Instruction Set Manual, Revision 6.06
DIV
Divide Word
ready. Asynchronous execution does not affect the program result, but offers an opportunity for performance improvement by scheduling the divide so that other instructions can execute in parallel.
Historical Perspective:
In MIPS 1 through MIPS III, if either of the two instructions preceding the divide is an MFHI or MFLO, the result of the MFHI or MFLO is UNPREDICTABLE. Reads of the HI or LO special register must be separated from subse- quent instructions that write to them by two or more instructions. This restriction was removed in MIPS IV and MIPS32 and all subsequent levels of the architecture.

DIV IDivide Word
The MIPS32® Instruction Set Manual, Revision 6.06 167

DIV MOD DIVU MODU Divide Integers (with result to GPR)
31
26 25 21 20 16 15 11 10 6 5 0
SPECIAL 000000
rs
rt
rd
DIV 00010
SOP32 011010
SPECIAL 000000
rs
rt
rd
MOD 00011
SOP32 011010
SPECIAL 000000
rs
rt
rd
DIVU 00010
SOP33 011011
SPECIAL 000000
rs
rt
rd
MODU 00011
SOP33 011011
168
The MIPS32® Instruction Set Manual, Revision 6.06
655556
Format: DIV MOD DIVU MODU DIV rd,rs,rt
MOD rd,rs,rt
DIVU rd,rs,rt
MODU rd,rs,rt
Purpose: Divide Integers (with result to GPR)
DIV: Divide Words Signed
MOD: Modulo Words Signed DIVU: Divide Words Unsigned MODU: Modulo Words Unsigned
Description:
DIV: GPR[rd]  ( divide.signed( GPR[rs], GPR[rt] ) MOD: GPR[rd]  ( modulo.signed( GPR[rs], GPR[rt] ) DIVU: GPR[rd]  ( divide.unsigned( GPR[rs], GPR[rt] ) MODU: GPR[rd]  ( modulo.unsigned( GPR[rs], GPR[rt] )
MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6
The Release 6 divide and modulo instructions divide the operands in GPR rs and GPR rt, and place the quotient or remainder in GPR rd.
For each of the div/mod operator pairs DIV/M OD, DIVU/MODU, the results satisfy the equation (AdivB)*B+(AmodB)=A, where (AmodB) has same sign as the dividend A, and abs(A mod B) < abs(B). This equation uniquely defines the results. NOTE: if the divisor B=0, this equation cannot be satisfied, and the result is UNPREDICTABLE. This is commonly called “truncated division”. DIV performs a signed 32-bit integer division, and places the 32-bit quotient result in the destination register. MOD performs a signed 32-bit integer division, and places the 32-bit remainder result in the destination register. The remainder result has the same sign as the dividend. DIVU performs an unsigned 32-bit integer division, and places the 32-bit quotient result in the destination register. MODU performs an unsigned 32-bit integer division, and places the 32-bit remainder result in the destination regis- ter. Restrictions: If the divisor in GPR rt is zero, the result value is UNPREDICTABLE. DIVMODDIVUMODU IDIV:DivideWordsSignedMOD:ModuloWordsSignedDIVU:DivideWordsUn- Availability and Compatibility: These instructions are introduced by and required as of Release 6. Release 6 divide instructions have the same opcode mnemonic as the pre-Release 6 divide instructions (DIV, DIVU). The instruction encodings are different, as are the instruction semantics: the Release 6 instruction produces only the quotient, whereas the pre-Release 6 instruction produces quotient and remainder in HI/LO registers respectively, and separate modulo instructions are required to obtain the remainder. The assembly syntax distinguishes the Release 6 from the pre-Release 6 divide instructions. For example, Release 6 “DIV rd,rs,rt” specifies 3 register operands, versus pre-Release 6 “DIV rs,rt”, which has only two register arguments, with the HI/LO registers implied. Some assemblers accept the pseudo-instruction syntax “DIV rd,rs,rt” and expand it to do “DIV rs,rt;MFHI rd”. Phrases such as “DIV with GPR output” and “DIV with HI/LO output” may be used when disambiguation is necessary. Pre-Release 6 divide instructions that produce quotient and remainder in the HI/LO registers produce a Reserved Instruction exception on Release 6. In the future, the instruction encoding may be reused for other instructions. Programming Notes: Because the divide and modulo instructions are defined to not trap if dividing by zero, it is safe to emit code that checks for zero-divide after the divide or modulo instruction. Operation DIV, MOD: s1 signed_word(GPR[rs]) s2 signed_word(GPR[rt]) DIVU, MODU: s1 unsigned_word(GPR[rs]) s2 unsigned_word(GPR[rt]) DIV, DIVU: quotient  s1 div s2 MOD, MODU: remainder  s1 mod s2 DIV: GPR[rd]  quotient MOD: GPR[rd]  remainder DIVU: GPR[rd]  quotient MODU: GPR[rd]  remainder /* end of instruction */ Exceptions: No arithmetic exceptions occur. Division by zero produces an UNPREDICTABLE result. The MIPS32® Instruction Set Manual, Revision 6.06 169 DIV.fmt IFloating Point Divide 31 26 25 21 20 16 15 11 10 6 5 655556 Format: DIV.fmt DIV.S fd, fs, ft DIV.D fd, fs, ft Purpose: Floating Point Divide To divide FP values. 0 MIPS32 MIPS32 COP1 010001 fmt ft fs fd DIV 000011 Description: FPR[fd]  FPR[fs] / FPR[ft] The value in FPR fs is divided by the value in FPR ft. The result is calculated to infinite precision, rounded according to the current rounding mode in FCSR, and placed into FPR fd. The operands and result are values in format fmt. Restrictions: The fields fs, ft, and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPREDICABLE. The operands must be values in format fmt; if they are not, the result is UNPREDICTABLE and the value of the operand FPRs becomes UNPREDICTABLE. Operation: StoreFPR (fd, fmt, ValueFPR(fs, fmt) / ValueFPR(ft, fmt)) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Inexact, Invalid Operation, Unimplemented Operation, Division-by-zero, Overflow, Underflow The MIPS32® Instruction Set Manual, Revision 6.06 170 DIVU IDivide Unsigned Word 31 26 25 21 20 16 15 6 5 0 6 5 5 10 6 Format: DIVU rs, rt MIPS32, removed in Release 6 Purpose: Divide Unsigned Word To divide 32-bit unsigned integers Description: (HI, LO)  GPR[rs] / GPR[rt] The 32-bit word value in GPR rs is divided by the 32-bit value in GPR rt, treating both operands as unsigned values. The 32-bit quotient is placed into special register LO and the 32-bit remainder is placed into special register HI. No arithmetic exception occurs under any circumstances. Restrictions: If the divisor in GPR rt is zero, the arithmetic result value is UNPREDICTABLE. Availability and Compatibility: This instruction has been removed in Release 6. Operation: q  (0 || GPR[rs]31..0) div (0 || GPR[rt]31..0) r  (0 || GPR[rs]31..0) mod (0 || GPR[rt]31..0) LO  sign_extend(q31..0) HI  sign_extend(r31..0) Exceptions: None Programming Notes: Pre-Release 6 instruction DIV has been removed in Release 6 and has been replaced by DIV and MOD instructions that produce only quotient and remainder, respectively. Refer to the Release 6 introduced ‘DIV’ and ‘MOD’ instruc- tions in this manual for more information. This instruction remains current for all release levels lower than Release 6 of the MIPS architecture. See “Programming Notes” for the DIV instruction. Historical Perspective: In MIPS 1 through MIPS III, if either of the two instructions preceding the divide is an MFHI or MFLO, the result of the MFHI or MFLO is UNPREDICTABLE. Reads of the HI or LO special register must be separated from subse- quent instructions that write to them by two or more instructions. This restriction was removed in MIPS IV and MIPS32 and all subsequent levels of the architecture. SPECIAL 000000 rs rt 0 00 0000 0000 DIVU 011011 The MIPS32® Instruction Set Manual, Revision 6.06 171 DVP IDisable Virtual Processor 31 26 25 21 20 16 15 11 10 6 5 4 3 2 0 65555123 Format: DVP rt MIPS32 Release 6 Purpose: Disable Virtual Processor To disable all virtual processors in a physical core other than the virtual processor that issued the instruction. Description: GPR[rt]  VPControl ; VPControlDIS  1 Disabling a virtual processor means that instruction fetch is terminated, and all outstanding instructions for the affected virtual processor(s) must be complete before the DVP itself is allowed to retire. Any outstanding events such as hardware instruction or data prefetch, or page-table walks must also be terminated. The DVP instruction has implicit SYNC(stype=0) semantics but with respect to the other virtual processors in the physical core. After all other virtual processors have been disabled, VPControlDIS is set. Prior to modification and if rt is non- zero, VPControl is written to GPR[rt].If DVP is specified without rt, then rtmust be 0. DVP may also take effect on a virtual processor that has executed a WAIT or a PAUSE instruction. If a virtual proces- sor has executed a WAIT instruction, then it cannot resume execution on an interrupt until an EVP has been executed. If the EVP is executed before the interrupt arrives, then the virtual processor resumes in a state as if the DVP had not been executed, that is, it waits for the interrupt. If a virtual processor has executed a PAUSE instruction, then it cannot resume execution until an EVP has been exe- cuted, even if LLbit is cleared. If an EVP is executed before the LLbit is cleared, then the virtual processor resumes in a state as if the DVP has not been executed, that is, it waits for the LLbit to clear. The execution of a DVP must be followed by the execution of an EVP. The execution of an EVP causes execution to resume immediately—where applicable—on all other virtual processors, as if the DVP had not been executed. The execution is completely restorable after the EVP. If an event occurs in between the DVP and EVP that renders state of the virtual processor UNPREDICTABLE (such as power-gating), then the effect of EVP is UNPREDICTABLE. DVP may only take effect if VPControlDIS=0. Otherwise it is treated as a NOP instruction. If a virtual processor is disabled due to a DVP, then interrupts are also disabled for the virtual processor, that is, logi- cally StatusIE=0. StatusIE for the target virtual processors though is not cleared though as software cannot access state on the virtual processors that have been disabled. Similarly, deferred exceptions will not cause a disabled virtual processor to be re-enabled for execution, at least until execution is re-enabled by the EVP instruction. The vir- tual processor that executes the DVP, however, continues to be interruptible. In an implementation, the ability of a virtual processor to execute instructions may also be under control external to the physical core which contains the virtual processor. If disabled by DVP, a virtual processor must not resume fetch in response to the assertion of this external signal to enable fetch. Conversely, if fetch is disabled by such external control, then execution of EVP will not cause fetch to resume at a target virtual processor for which the control is deasserted. This instruction never executes speculatively. It must be the oldest unretired instruction to take effect. This instruction is only available in Release 6 implementations. For implementations that do not support multi- threading (Config5VP=0), this instruction must be treated as a NOP instruction. Restrictions: If access to Coprocessor 0 is not enabled, a Coprocessor Unusable Exception is signaled. COP0 010000 MFMC0 01011 rt 0 00000 0 00000 sc 1 0 00 4 100 The MIPS32® Instruction Set Manual, Revision 6.06 172 173 The MIPS32® Instruction Set Manual, Revision 6.06 DVP Disable Virtual Processor In implementations prior to Release 6 of the architecture, this instruction resulted in a Reserved Instruction exception. Operation: The pseudo-code below assumes that the DVP is executed by virtual processor 0, while the target virtual processor is numbered ‘n’, where n is each of all remaining virtual processors. if (VPControlDIS = 0) // Pseudo-code in italics provides recommended action wrt other VPs disable_fetch(VPn) { if PAUSE(VPn) retires prior or at disable event then VPn execution is not resumed if LLbit is cleared prior to EVP } disable_interrupt(VPn) { if WAIT(VPn) retires prior or at disable event then interrupts are ignored by VPn until EVP } // DVP0 not retired until instructions for VPn completed while (VPn outstanding instruction) DVP0 unretired endwhile endif data  VPControl GPR[rt]  data VPControlDIS  1 Exceptions: Coprocessor Unusable Reserved Instruction (pre-Release 6 implementations) Programming Notes: DVP may disable execution in the target virtual processor regardless of the operating mode - kernel, supervisor, user. Kernel software may also be in a critical region, or in a high-priority interrupt handler when the disable occurs. Since the instruction is itself privileged, such events are considered acceptable. Before executing an EVP in a DVP/EVP pair, software should first read VPControlDIS, returned by DVP, to deter- mine whether the virtual processors are already disabled. If so, the DVP/EVP sequence should be abandoned. This step allows software to safely nest DVP/EVP pairs. Privileged software may use DVP/EVP to disable virtual processors on a core, such as for the purpose of doing a cache flush without interference from other processes in a system with multiple virtual processors or physical cores. DVP (and EVP) may be used in other cases such as for power-savings or changing state that is applicable to all virtual processors in a core, such as virtual processor scheduling priority, as described below: ll t0 0(a0) dvp // disable all other virtual processors pause // wait for LLbit to clear evp // enable all othe virtual processors DVP IDisable Virtual Processor ll t0 0(a0) dvp // disable all other virtual processors
evp // enable all othe virtual processors
The MIPS32® Instruction Set Manual, Revision 6.06
174

EHB
Execution Hazard Barrier
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: EHB AssemblyIdiomMIPS32Release2 Purpose: Execution Hazard Barrier
To stop instruction execution until all execution hazards have been cleared.
Description:
EHB is used to denote execution hazard barrier. The actual instruction is interpreted by the hardware as SLL r0, r0, 3.
This instruction alters the instruction issue behavior on a pipelined processor by stopping execution until all execu- tion hazards have been cleared. Other than those that might be created as a consequence of setting StatusCU0, there are no execution hazards visible to an unprivileged program running in User Mode. All execution hazards created by previous instructions are cleared for instructions executed immediately following the EHB, even if the EHB is exe- cuted in the delay slot of a branch or jump. The EHB instruction does not clear instruction hazards—such hazards are cleared by the JALR.HB, JR.HB, and ERET instructions.
Restrictions:
None
Operation:
ClearExecutionHazards()
Exceptions:
None
Programming Notes:
In Release 2 implementations, this instruction resolves all execution hazards. On a superscalar processor, EHB alters the instruction issue behavior in a manner identical to SSNOP. For backward compatibility with Release 1 implemen- tations, the last of a sequence of SSNOPs can be replaced by an EHB. In Release 1 implementations, the EHB will be treated as an SSNOP, thereby preserving the semantics of the sequence. In Release 2 implementations, replacing the final SSNOP with an EHB should have no performance effect because a properly sized sequence of SSNOPs will have already cleared the hazard. As EHB becomes the standard in MIPS implementations, the previous SSNOPs can be removed, leaving only the EHB.
SPECIAL 000000
0 00000
0 00000
0 00000
3 00011
SLL 000000
175
The MIPS32® Instruction Set Manual, Revision 6.06

EI
IEnable Interrupts
31
26 25 21 20 16 15 11 10 6 5 4 3 2 0
65555123
Format: EI MIPS32Release2 EI rt MIPS32 Release 2
Purpose: Enable Interrupts
To return the previous value of the Status register and enable interrupts. If EI is specified without an argument, GPR
r0 is implied, which discards the previous value of the Status register. Description: GPR[rt]  Status; StatusIE  1
The current value of the Status register is loaded into general register rt. The Interrupt Enable (IE) bit in the Status register is then set.
Restrictions:
If access to Coprocessor 0 is not enabled, a Coprocessor Unusable Exception is signaled.
In implementations prior to Release 2 of the architecture, this instruction resulted in a Reserved Instruction exception.
Operation:
This operation specification is for the general interrupt enable/disable operation, with the sc field as a variable. The individual instructions DI and EI have a specific value for the sc field.
data  Status GPR[rt]  data StatusIE  1
Exceptions:
Coprocessor Unusable
Reserved Instruction (Release 1 implementations)
Programming Notes:
The effects of this instruction are identical to those accomplished by the sequence of reading Status into a GPR, set- ting the IE bit, and writing the result back to Status. Unlike the multiple instruction sequence, however, the EI instruction cannot be aborted in the middle by an interrupt or exception.
This instruction creates an execution hazard between the change to the Status register and the point where the change to the interrupt enable takes effect. This hazard is cleared by the EHB, JALR.HB, JR.HB, or ERET instructions. Soft- ware must not assume that a fixed latency will clear the execution hazard.
COP0 0100 00
MFMC0 01 011
rt
12 0110 0
0 000 00
sc 1
0 00
0 000
The MIPS32® Instruction Set Manual, Revision 6.06 176

ERET
IException Return
31
26 25 24 6 5 0
COP0 010000
CO 1
0
000 0000 0000 0000 0000
ERET 011000
61 19 6
Format: ERET
Purpose: Exception Return
To return from interrupt, exception, or error trap.
Description:
MIPS32
ERET clears execution and instruction hazards, conditionally restores SRSCtlCSS from SRSCtlPSS in a Release 2 implementation, and returns to the interrupted instruction at the completion of interrupt, exception, or error process- ing. ERET does not execute the next instruction (that is, it has no delay slot).
Restrictions:
Pre-Release 6: The operation of the processor is UNDEFINED if an ERET is executed in the delay slot of a branch
or jump instruction.
Release 6: Implementations are required to signal a Reserved Instruction exception if ERET is encountered in the delay slot or forbidden slot of a branch or jump instruction.
An ERET placed between an LL and SC instruction will always cause the SC to fail.
ERET implements a software barrier that resolves all execution and instruction hazards created by Coprocessor 0 state changes (for Release 2 implementations, refer to the SYNCI instruction for additional information on resolving instruction hazards created by writing the instruction stream). The effects of this barrier are seen starting with the instruction fetch and decode of the instruction at the PC to which the ERET returns.
In a Release 2 implementation, ERET does not restore SRSCtlCSS from SRSCtlPSS if StatusBEV = 1, or if StatusERL = 1 because any exception that sets StatusERL to 1 (Reset, Soft Reset, NMI, or cache error) does not save SRSCtlCSS in SRSCtlPSS. If software sets StatusERL to 1, it must be aware of the operation of an ERET that may be subse- quently executed.
Operation:
if StatusERL = 1 then temp  ErrorEPC StatusERL  0
else
temp  EPC
StatusEXL  0
if (ArchitectureRevision() ≥ 2) and (SRSCtlHSS  0) and (StatusBEV = 0) then
SRSCtlCSS  SRSCtlPSS endif
endif
if IsMIPS16Implemented() | (Config3ISA  0) then
PC  temp31..1 || 0
ISAMode  temp0 else
PC  temp endif
LLbit  0 ClearHazards()
The MIPS32® Instruction Set Manual, Revision 6.06 177

ERET IException Return Exceptions:
Coprocessor Unusable Exception
The MIPS32® Instruction Set Manual, Revision 6.06 178

ERETNC IException Return No Clear
31
26 25 24 6 5 0
61 18 16
Format: ERETNC MIPS32Release5 Purpose: Exception Return No Clear
To return from interrupt, exception, or error trap without clearing the LLbit.
Description:
ERETNC clears execution and instruction hazards, conditionally restores SRSCtlCSS from SRSCtlPSS when imple- mented, and returns to the interrupted instruction at the completion of interrupt, exception, or error processing. ERETNC does not execute the next instruction (i.e., it has no delay slot).
ERETNC is identical to ERET except that an ERETNC will not clear the LLbit that is set by execution of an LL instruction, and thus when placed between an LL and SC sequence, will never cause the SC to fail.
An ERET must continue to be used by default in interrupt and exception processing handlers. The handler may have accessed a synchronizable block of memory common to code that is atomically accessing the memory, and where the code caused the exception or was interrupted. Similarly, a process context-swap must also continue to use an ERET in order to avoid a possible false success on execution of SC in the restored context.
Multiprocessor systems with non-coherent cores (i.e., without hardware coherence snooping) should also continue to use ERET, because it is the responsibility of software to maintain data coherence in the system.
An ERETNC is useful in cases where interrupt/exception handlers and kernel code involved in a process context- swap can guarantee no interference in accessing synchronizable memory across different contexts. ERETNC can also be used in an OS-level debugger to single-step through code for debug purposes, avoiding the false clearing of the LLbit and thus failure of an LL and SC sequence in single-stepped code.
Software can detect the presence of ERETNC by reading Config5LLB. Restrictions:
Release 6 implementations are required to signal a Reserved Instruction exception if ERETNC is executed in the delay slot or Release 6 forbidden slot of a branch or jump instruction.
ERETNC implements a software barrier that resolves all execution and instruction hazards created by Coprocessor 0 state changes. (For Release 2 implementations, refer to the SYNCI instruction for additional information on resolving instruction hazards created by writing the instruction stream.) The effects of this barrier are seen starting with the instruction fetch and decode of the instruction in the PC to which the ERETNC returns.
Operation:
if StatusERL = 1 then temp  ErrorEPC StatusERL  0
else
temp  EPC
StatusEXL  0
if (ArchitectureRevision() ≥ 2) and (SRSCtlHSS  0) and (StatusBEV = 0) then
SRSCtlCSS  SRSCtlPSS endif
endif
if IsMIPS16Implemented() | (Config3ISA  0) then
COP0 010000
CO 1
0
000 0000 0000 0000 000
1
ERET 011000
The MIPS32® Instruction Set Manual, Revision 6.06 179

ERETNC
IException Return No Clear
PC  temp31..1 || 0
ISAMode  temp0 else
PC  temp endif
ClearHazards()
Exceptions:
Coprocessor Unusable Exception
The MIPS32® Instruction Set Manual, Revision 6.06
180

EVP
31
IEnable Virtual Processor 26 25 21 20 16 15 11 10 6 5 4 3 2 0
65555123
Format: EVPrt MIPS32Release6 Purpose: Enable Virtual Processor
To enable all virtual processors in a physical core other than the virtual processor that issued the instruction.
Description: GPR[rt]  VPControl ; VPControlDIS  0 Enabling a virtual processor means that instruction fetch is resumed.
After all other virtual processors have been enabled, VPControlDIS is cleared. Prior to modification, if rt is non- zero, VPControl is written to GPR[rt].If EVP is specified without rt, then rtmust be 0.
See the DVP instruction to understand the application of EVP in the context of WAIT/PAUSE/external-control (“DVP” on page 172).
The execution of a DVP must be followed by the execution of an EVP. The execution of an EVP causes execution to resume immediately, where applicable, on all other virtual processors, as if the DVP had not been executed, that is, execution is completely restorable after the EVP. On the other hand, if an event occurs in between the DVP and EVP that renders state of the virtual processor UNPREDICTABLE (such as power-gating), then the effect of EVP is UNPREDICTABLE.
EVP may only take effect if VPControlDIS=1. Otherwise it is treated as a NOP
This instruction never executes speculatively. It must be the oldest unretired instruction to take effect.
This instruction is only available in Release 6 implementations. For implementations that do not support multi- threading (Config5VP=0), this instruction must be treated as a NOP instruction.
Restrictions:
If access to Coprocessor 0 is not enabled, a Coprocessor Unusable Exception is signaled.
In implementations prior to Release 6 of the architecture, this instruction resulted in a Reserved Instruction exception.
Operation:
The pseudo-code below assumes that the EVP is executed by virtual processor 0, while the target virtual processor is numbered ‘n’, where n is each of all remaining virtual processors.
if (VPControlDIS = 1)
// Pseudo-code in italics provides recommended action wrt other VPs
enable_fetch(VPn) {
if PAUSE(VPn) retires prior or at disable event
then VPn execution is not resumed if LLbit is cleared prior to EVP
}
enable_interrupt(VPn) {
if WAIT(VPn) retires prior or at disable event
then interrupts are ignored by VPn until EVP
}
COP0 010000
MFMC0 01011
rt
0 00000
0 00000
sc 0
0 00
4 100
The MIPS32® Instruction Set Manual, Revision 6.06 181

182
The MIPS32® Instruction Set Manual, Revision 6.06
EVP
Enable Virtual Processor
endif
data  VPControl GPR[rt]  data VPControlDIS  0
Exceptions:
Coprocessor Unusable
Reserved Instruction (pre-Release 6 implementations)
Programming Notes:
Before executing an EVP in a DVP/EVP pair, software should first read VPControlDIS, returned by DVP, to deter- mine whether the virtual processors are already disabled. If so, the DVP/EVP sequence should be abandoned. This step allows software to safely nest DVP/EVP pairs.
Privileged software may use DVP/EVP to disable virtual processors on a core, such as for the purpose of doing a cache flush without interference from other processes in a system with multiple virtual processors or physical cores.
DVP (and EVP) may be used in other cases such as for power-savings or changing state that is applicable to all virtual processors in a core, such as virtual processor scheduling priority, as described below:
ll t0 0(a0)
dvp // disable all other virtual processors pause // wait for LLbit to clear
evp // enable all othe virtual processors
ll t0 0(a0)
dvp // disable all other virtual processors
evp // enable all othe virtual processors

184
The MIPS32® Instruction Set Manual, Revision 6.06
EXT
Extract Bit Field
Exceptions:
Reserved Instruction

FLOOR.L.fmt IFloating Point Floor Convert to Long Fixed Point
31
26 25 21 20 16 15 11 10 6 5 0
655556
COP1 010001
fmt
0 00000
fs
fd
FLOOR.L 001011
Format: FLOOR.L.fmt FLOOR.L.S fd, fs
FLOOR.L.D fd, fs
Purpose: Floating Point Floor Convert to Long Fixed Point To convert an FP value to 64-bit fixed point, rounding down
MIPS32 Release 2 MIPS32 Release 2
Description: FPR[fd]  convert_and_round(FPR[fs])
The value in FPR fs, in format fmt, is converted to a value in 64-bit long fixed point format and rounded toward 
(rounding mode 3). The result is placed in FPR fd.
When the source value is Infinity, NaN, or rounds to an integer outside the range -263 to 263-1, the result cannot be represented correctly, an IEEE Invalid Operation condition exists, and the Invalid Operation flag is set in the FCSR. If the Invalid Operation Enable bit is set in the FCSR, no result is written to fd and an Invalid Operation exception is taken immediately. Otherwise, a default result is written to fd. On cores with FCSRNAN2008=0, the default result is
263–1. On cores with FCSRNAN2008=1, the default result is:
• 0 when the input value is NaN
• 263–1 when the input value is + or rounds to a number larger than 263–1
• -263–1 when the input value is – or rounds to a number smaller than -263–1
Restrictions:
The fields fs and fd must specify valid FPRs: fs for type fmt and fd for long fixed point. If the fields are not valid, the result is UNPREDICTABLE.
The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE.
The result of this instruction is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model; it is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU.
Operation:
StoreFPR(fd, L, ConvertFmt(ValueFPR(fs, fmt), fmt, L))
Exceptions:
Coprocessor Unusable, Reserved Instruction
Floating Point Exceptions:
Invalid Operation, Unimplemented Operation, Inexact
The MIPS32® Instruction Set Manual, Revision 6.06 185

FLOOR.W.fmt Floating Point Floor Convert to Word Fixed Point
31
26 25 21 20 16 15 11 10 6 5
655556
Format: FLOOR.W.fmt FLOOR.W.S fd, fs
FLOOR.W.D fd, fs
Purpose: Floating Point Floor Convert to Word Fixed Point To convert an FP value to 32-bit fixed point, rounding down
0
MIPS32 MIPS32
COP1 010001
fmt
0 00000
fs
fd
FLOOR.W 001111
186
The MIPS32® Instruction Set Manual, Revision 6.06
Description: FPR[fd]  convert_and_round(FPR[fs])
The value in FPR fs, in format fmt, is converted to a value in 32-bit word fixed point format and rounded toward –
(rounding mode 3). The result is placed in FPR fd.
When the source value is Infinity, NaN, or rounds to an integer outside the range -231 to 231-1, the result cannot be represented correctly, an IEEE Invalid Operation condition exists, and the Invalid Operation flag is set in the FCSR. If the Invalid Operation Enable bit is set in the FCSR, no result is written to fd and an Invalid Operation exception is taken immediately. Otherwise, a default result is written to fd. On cores with FCSRNAN2008=0, the default result is
231–1. On cores with FCSRNAN2008=1, the default result is:
• 0 when the input value is NaN
• 231–1 when the input value is + or rounds to a number larger than 231–1
• -231–1 when the input value is – or rounds to a number smaller than -231–1
Restrictions:
The fields fs and fd must specify valid FPRs: fs for type fmt and fd for word fixed point. If the fields are not valid, the result is UNPREDICTABLE.
The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE.
Operation:
StoreFPR(fd, W, ConvertFmt(ValueFPR(fs, fmt), fmt, W))
Exceptions:
Coprocessor Unusable, Reserved Instruction
Floating Point Exceptions:
Invalid Operation, Unimplemented Operation, Inexact

GINVI IGlobal Invalidate Instruction Cache
31
26 25 21 20 8 7 6 5 0
SPECIAL3 011111
rs
0
0000000000000
GINVI 00
GINV 111101
6 5 13 2
6
Format: GINVI rs MIPS32 Release 6 Purpose: Global Invalidate Instruction Cache
In a multi-processor system, fully invalidate all remote primary instruction-caches, or a specified single cache. The local primary instruction cache is also fully invalidated in the case where all the remote caches are to be invalidated.
Description: Invalidate_All_Primary_Instruction_Caches(null or rs)
Fully invalidate all remote primary instruction caches, or a specified single cache, whether local or remote. The local
primary instruction cache is also fully invalidated in the case where all remote caches are to be invalidated.
If rs field of the opcode is 0, then all caches are to be invalidated. ‘rs’ should be specified as 0 in the assembly syntax for this case. If rs field of the opcode is not 0, then a single cache that is specified by an implementation dependent number of lower bits of GPR[rs] is invalidated, which may be the local cache itself.
Software based invalidation of the primary instruction cache is required in a system if coherency of the cache is not maintained in hardware. While typically limited to the primary cache, the scope of the invalidation within a processor is however implementation dependent – it should apply to all instruction caches within the cache hierarchy that required software coherence maintenance.
In legacy systems, it is software’s responsibility to keep the instruction cache state consistent through SYNCI instruc- tions. This instruction provides a method for bulk invalidating the instruction caches in lieu of SYNCI.
The instruction’s action is considered complete when the both the local and remote cache invalidations are complete, that is, the data in the cache is no longer available to the related instruction stream. Whether these invalidations are complete can only be determined by the completion of a SYNC (stype=0x14) that follows the invalidate instruc- tion(s). With the completion of the SYNC operation, all global invalidations preceding the SYNC in the program are considered globally visible.
Whether the SYNC(stype=0x14) or the global invalidate itself cause synchronization of the instruction stream to new state/context is implementation dependent.
A processor may send a global invalidate instruction remotely only when any preceding global invalidate for the pro- gram has reached a global ordering point.
The GINVI has no instruction or execution hazard barrier semantics in itself.
If the implementation allows a cache line to be locked, i.e., not replaceable during a fill, GINVI will not invalidate the line. A cache line can be locked through the optional CACHE “Fetch and Lock” instruction.
See Programming Notes for programming constraints.
Restrictions:
If an implementation does not support the instruction, a Reserved Instruction exception is caused. If access to Coprocessor 0 is not enabled, a Coprocessor Unusable Exception is signaled.
In a single processor SOC, this instruction acts on the local instruction cache only.
Operation:
Local:
if (Config5GI ≠ 2’b1x)
The MIPS32® Instruction Set Manual, Revision 6.06 187
Copyright © 2016 MIPS Technologies Inc. All rights reserved.

188
The MIPS32® Instruction Set Manual, Revision 6.06
GINVI
Global Invalidate Instruction Cache
SignalException(ReservedInstruction) // if not implemented
break endif
if IsCoprocessorEnabled(0) then
// Fully invalidate local instruction cache, if selected.
// Send invalidation message to other cores, if required.
else
SignalException(CoprocessorUnusable, 0)
endif
Remote:
// Fully invalidate remote instruction cache.
Exceptions:
Reserved Instruction, Coprocessor Unusable
Programming Notes:
For the local processor, the instruction stream is synchronized by an instruction hazard barrier such as JR.HB.
The instruction stream in the remote processor is synchronized with respect to the execution of GINVI once the SYNC operation following GINVI completes.
The following sequence is recommended for use of GINVI.
ginvi /* fully-invalidate all caches*/
sync 0x14 /* Enforce completion – all instruction streams synchronized. */
jr.hb ra /* Clear instruction hazards*/
Implementation Notes:
None.
Copyright © 2016 MIPS Technologies Inc. All rights reserved.

GINVT IGlobal Invalidate TLB
31
26 25 21 20 16 15 10 9 8 7 6 5
655622
Format: GINVT rs, type Purpose: Global Invalidate TLB
In a multi-processor system, invalidate translations of remote TLBs and local TLB.
Description: Invalidate_TLB(GPR[rs], MemoryMapID)
0
SPECIAL3 011111
rs
00000
000000
type
GINVT 10
GINV 111101
Invalidate a TLB in multiple ways – entire TLB, by Virtual Address and MemoryMapID, by MemoryMapID or Vir- tual Address. The Virtual Address is obtained from GPR[rs]. The MemoryMapID is derived from CP0 MemoryMapID. The virtual address is associated with a specific Memory Map identified by MemoryMapID.
The virtual address within GPR[rs] is aligned to twice the size of the minimum page size of 4KB i.e., it is equivalent to EntryHiVPN2: bit 13 of the virtual address is aligned to bit 13 of GPR[rs]. If the virtual address is not required, such as in the case of invalidate All or by MemoryMapID, then ‘rs’ should be specified as 0 in the assembly syntax but is otherwise ignored.
The MemoryMapID is a replacement for EntryHiASID. The MemoryMapID is an implementation-dependent num- ber of bits that must be larger than the existing EntryHiASID (10-bits including EntryHiASIDX ). The purpose of
a larger tag is to be able to uniquely identify processes in the system. A 16-bit MemoryMapID for example will iden- tify 64K Memory Maps, while the current 8-bit ASID only identifies 256, and is thus subject to frequent recycling. An implementation with MemoryMapID is designed to be backward compatible with software that uses EntryHiA- SID. See CP0 MemoryMapID.
Table 3.1 specifies the different types of invalidates supported as a function of the “type” field of the instruction. Table 3.1 Types of Global TLB Invalidates
With reference to Table 3.1, if the Global bit in a TLB entry is set, then MemoryMapID comparison is ignored by the operation.
The instruction is considered complete when the local and remote invalidations are complete. Whether these invalida- tions are complete can only be determined by the completion of a SYNC (stype=0x14) that follows the invalidate instruction(s). With the completion of the SYNC operation, all invalidations of this type preceding the SYNC in the program are considered globally visible.
Whether the SYNC(stype=0x14) or the global invalidate itself cause synchronization of the instruction stream to new state/context is implementation dependent.
A GINVT based invalidation is complete, whether local or remote, when the following has occurred: the TLB is invalidated of matching entries, and all instructions in the instruction stream after the point of completion can only access the new context.
6
MIPS32 Release 6
Encoding of “type” field
Definition
00
Invalidate entire TLB
01
Invalidate by VA (MemoryMapID is globalized)
10
Invalidate by MemoryMapID
11
Invalidate by VA and MemoryMapID.
The MIPS32® Instruction Set Manual, Revision 6.06 189
Copyright © 2016 MIPS Technologies Inc. All rights reserved.

190
A processor may send a global invalidate instruction remotely only when any preceding global invalidate for the pro- gram has reached a global ordering point.
GINVT has no instruction or execution hazard barrier semantics in itself.
A GINVT operation that is specified to invalidate all entries will only invalidate non-wired entries. Other GINVT operations will invalidate wired entries on a match.
Restrictions:
If an implementation does not support the instruction, or use of MemoryMapID is disabled, then a Reserved Instruc- tion exception is caused.
If access to Coprocessor 0 is not enabled, a Coprocessor Unusable exception is signaled. In a single processor SOC, this instruction acts on the local TLB only.
Operation:
Local:
if (Config5GI ≠ 2’b11) then
SignalException(ReservedInstruction, 0)
break endif
if IsCoprocessorEnabled(0) then if (Config5MI = 1)
// generate control from instruction encoding. invAll  (ginvt[type] = 0b00)
invVA  (ginvt[type] = 0b01)
invMMid  (ginvt[type] = 0b10)
invVAMMid  (ginvt[type] = 0b11)
// generate data; how data is driven when unsupported is imp-dep. // Format of GPR[rs] equals CP0 EntryHi.
InvMsgVPN2 GPR[rs]31..13
InvMsgMMid  MemoryMapID // imp-dep # of bits
// Broadcast invalidation message to other cores. InvalidateTLB(InvMsgVPN2,InvMsgMMid,invAll,invVAMMid,invMMid,invVA)
else // if not implemented, MMid disabled
SignalException(ReservedInstruction)
endif else
SignalException(CoprocessorUnusable, 0)
endif
Remote:
// Repeat in all remote TLBs InvalidateTLB(InvMsgVPN2,InvMsgMMid,invAll,invVAMMid,invMMid,invVA)
function InvalidateTLB(InvMsgVPN2,InvMsgMMid,invAll,invVAMMid,invMMid,invVA) // “Mask” is equivalent to CP0 PageMask.
// “G” is equivalent to the Global bit in CP0 EntryLo0/1.
for i in 0..TLBEntries-1
// Wired entries are excluded.
VAMatch  ((TLB[i]VPN2 and not TLB[i]Mask) = (InvMsgVPN2 and not TLB[i]Mask))
MMidMatch  (TLB[i]MMid = InvMsgMMid)
if ((invAll and (i>CP0.Wired.Wired)) or // do not invalidate Wired (VAMatch and ((TLB[i]G = 1) or MMidMatch) and invVAMMid) or (VAMatch and invVA) or
The MIPS32® Instruction Set Manual, Revision 6.06
GINVT Global Invalidate TLB
Copyright © 2016 MIPS Technologies Inc. All rights reserved.

GINVT
IGlobal Invalidate TLB
(MMidMatch and (TLB[i]G ≠ 1) and invMMid)) then
TLB[i]HW_Valid  0 endif
endfor
endfunction
Exceptions:
Reserved Instruction, Coprocessor Unusable
Programming Notes:
// where HW_Valid is the entry valid bit
Since CP0 MemoryMapID sources the value of MemoryMapID of the currently running process, the kernel must save/ restore MemoryMapID appropriately before it modifies it for the invalidation operation. Between the save and restore, it must utilize unmapped addresses.
An MTC0 that modifies MemoryMapID must be followed by an EHB to make this value visible to a subsequent GINVT. Where multiple GINVTs are used prior to a single SYNC (stype=0x14), each may use a different value of MemoryMapID.
For the local processor, the instruction stream is synchronized to the new translation context (where applicable) by an instruction hazard barrier such as JR.HB.
The instruction stream in the remote processor is synchronized with respect to the execution of GINVT once the SYNC operation completes.
The following sequence is recommended for use of GINVT.
mtc0 0, C0_PWCtl /* disable Page Walker,where applicable;implementation-dependent*/
ehb
ginvt r1, type
sync 0x14
jr.hb ra
/* Clear execution hazards to prevent speculative walks*/
/* Invalidate TLB(s) */
/* Enforce completion */
/* Clear instruction hazards */
Whether the hardware page table walker, if implemented, needs to be disabled as shown above, is implementation dependent. It is recommended that hardware take the steps to locally disable the hardware page table walker to main- tain TLB consistency, as it would for remote TLBs.
Software must take into account a system that may have potentially varying widths of MemoryMapID . While not rec- ommended, different processors may have different implemented or programmed widths. Further, the interface between processors may support yet another width. If this is the case, then software responsible for global invalidates should be run on the processor with maximum width. Software must zero-fill any bits that are unused by a target. Software should also be able to rely on the implementation zero-filling bits where widths increase across any inter- face.
If an intermediate interface between source and target truncates the width of MemoryMapID, then software could address this limitation through various means: It could restrict the use of MemoryMapID to the interface width, it could program MemoryMapID with the expectation that over-invalidation may occur, or it should default to legacy means of invalidating the caches to prevent unreliable system behavior.
The MIPS32® Instruction Set Manual, Revision 6.06 191
Copyright © 2016 MIPS Technologies Inc. All rights reserved.

INS
Insert Bit Field
The operation is UNPREDICTABLE if lsb > msb. Operation:
if lsb > msb) then
UNPREDICTABLE
endif
GPR[rt]  GPR[rt]31..msb+1 || GPR[rs]msb-lsb..0 || GPR[rt]lsb-1..0
Exceptions:
Reserved Instruction
The MIPS32® Instruction Set Manual, Revision 6.06
193

J
Jump
31
26 25 0
6 26
Format: J target MIPS32 Purpose: Jump
To branch within the current 256 MB-aligned region.
Description:
This is a PC-region branch (not PC-relative); the effective target address is in the “current” 256 MB-aligned region. The low 28 bits of the target address is the instr_index field shifted left 2bits. The remaining upper bits are the corre- sponding bits of the address of the instruction in the delay slot (not the branch itself).
Jump to the effective target address. Execute the instruction that follows the jump, in the branch delay slot, before executing the jump itself.
Restrictions:
Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE.
Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump.
Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception.
J 000010
instr_index
194
The MIPS32® Instruction Set Manual, Revision 6.06
Operation:
I:
I+1: PC  PC
Exceptions:
None
|| instr_index || 02
Programming Notes:
GPRLEN-1..28
Forming the branch target address by catenating PC and index bits rather than adding a signed offset to the PC is an advantage if all program code addresses fit into a 256MB region aligned on a 256MB boundary. It allows a branch from anywhere in the region to anywhere in the region, an action not allowed by a signed relative offset.
This definition creates the following boundary case: When the jump instruction is in the last word of a 256MB region, it can branch only to the following 256MB region containing the branch delay slot.
The Jump instruction has been deprecated in Release 6. Use BC instead.

JAL
31
26 25
6 26
Format: JALtarget Purpose: Jump and Link
To execute a procedure call within the current 256MB-aligned region.
Description:
IJump and Link 0
MIPS32
JAL 000011
instr_index
Place the return address link in GPR 31. The return link is the address of the second instruction following the branch, at which location execution continues after a procedure call.
This is a PC-region branch (not PC-relative); the effective target address is in the “current” 256MB-aligned region. The low 28 bits of the target address is the instr_index field shifted left 2bits. The remaining upper bits are the corre- sponding bits of the address of the instruction in the delay slot (not the branch itself).
Jump to the effective target address. Execute the instruction that follows the jump, in the branch delay slot, before executing the jump itself.
Restrictions:
Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE.
Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump.
Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception.
Operation:
I: GPR[31]  PC + 8
I+1: PC  PC GPRLEN-1..28
Exceptions:
None
Programming Notes:
|| instr_index || 02
Forming the branch target address by catenating PC and index bits rather than adding a signed offset to the PC is an advantage if all program code addresses fit into a 256MB region aligned on a 256MB boundary. It allows a branch from anywhere in the region to anywhere in the region, an action not allowed by a signed relative offset.
This definition creates the following boundary case: When the branch instruction is in the last word of a 256MB region, it can branch only to the following 256MB region containing the branch delay slot.
The Jump-and-Link instruction has been deprecated in Release 6. Use BALC instead.
The MIPS32® Instruction Set Manual, Revision 6.06 195

JALR Jump and Link Register pre-Release 6
31 26 25 21 20 16 15 11 10 6 5 0
655556
SPECIAL 000000
rs
0 00000
rd
hint
JALR 001001
Release 6
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: JALR rs (rd = 31 implied) MIPS32 JALR rd, rs MIPS32
Purpose: Jump and Link Register
To execute a procedure call to an instruction address in a register
Description: GPR[rd]  return_addr, PC  GPR[rs]
Place the return address link in GPR rd. The return link is the address of the second instruction following the branch,
where execution continues after a procedure call.
For processors that do not implement the MIPS16e or microMIPS ISA:
• Jump to the effective target address in GPR rs. If the target address is not 4-byte aligned, an Address Error
exception will occur when the target address is fetched.
For processors that do implement the MIPS16e or microMIPS ISA:
• Jump to the effective target address in GPR rs. Set the ISA Mode bit to the value in GPR rs bit 0. Set bit 0 of the target address to zero. If the target ISA Mode bit is 0 and the target address is not 4-byte aligned, an Address Error exception will occur when the target instruction is fetched.
In both cases, execute the instruction that follows the jump, in the branch delay slot, before executing the jump itself.
In Release 1 of the architecture, the only defined hint field value is 0, which sets default handling of JALR. In Release 2 of the architecture, bit 10 of the hint field is used to encode a hazard barrier. See the JALR.HB instruction description for additional information.
Restrictions:
Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE.
Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump.
Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception.
Jump-and-Link Restartability: Register specifiers rs and rd must not be equal, because such an instruction does not have the same effect when re-executed. The result of executing such an instruction is UNPREDICTABLE. This restriction permits an exception handler to resume execution by re-executing the branch when an exception occurs in the delay slot.
Restrictions Related to Multiple Instruction Sets: This instruction can change the active instruction set, if more than one instruction set is implemented.
SPECIAL 000000
rs
0 00000
rd
rd  00000
hint
JALR 001001
196
The MIPS32® Instruction Set Manual, Revision 6.06

JALR
IJump and Link Register
If only one instruction set is implemented, then the effective target address must obey the alignment rules of the instruction set. If multiple instruction sets are implemented, the effective target address must obey the alignment rules of the intended instruction set of the target address as specified by the bit 0 or GPR rs.
For processors that do not implement the microMIPS32/64 ISA, the effective target address in GPR rs must be natu- rally-aligned. For processors that do not implement the MIPS16e ASE nor microMIPS32/64 ISA, if either of the two least-significant bits are not zero, an Address Error exception occurs when the branch target is subsequently fetched as an instruction.
For processors that do implement the MIPS16e ASE or microMIPS32/64 ISA, if target ISAMode bit is zero (GPR rs bit 0) and bit 1 is one, an Address Error exception occurs when the jump target is subsequently fetched as an instruc- tion.
Availability and Compatibility:
Release 6 maps JR and JR.HB to JALR and JALR.HB with rd = 0:
Pre-Release 6, JR and JALR were distinct instructions, both with primary opcode SPECIAL, but with distinct func-
tion codes.
Release 6: JR is defined to be JALR with the destination register specifier rd set to 0. The primary opcode and func- tion field are the same for JR and JALR. The pre-Release 6 instruction encoding for JR is removed in Release 6.
Release 6 assemblers should accept the JR and JR.HB mnemonics, mapping them to the Release 6 instruction encod- ings.
Operation:
I: temp  GPR[rs] GPR[rd]  PC + 8
I+1:if (Config3ISA = 0) and (Config1CA = 0) then PC  temp
else
PC  tempGPRLEN-1..1 || 0
ISAMode  temp0 endif
Exceptions:
None
Programming Notes:
This jump-and-link register instruction can select a register for the return link; other link instructions use GPR 31. The default register for GPR rd, if omitted in the assembly language instruction, is GPR 31.
The MIPS32® Instruction Set Manual, Revision 6.06 197

JALR.HB Jump and Link Register with Hazard Barrier pre-Release 6:
31 26 25 21 20 16 15 11 10 9 6 5 0
6555146
SPECIAL 000000
rs
0 00000
rd
1
Any other legal hint value
JALR 001001
Release 6:
31
26 25 21 20 16 15 11 10 9 6 5 0
6555146
Format: JALR.HB rs (rd = 31 implied) MIPS32 Release 2 JALR.HB rd, rs MIPS32 Release 2
Purpose: Jump and Link Register with Hazard Barrier
To execute a procedure call to an instruction address in a register and clear all execution and instruction hazards
Description: GPR[rd]  return_addr, PC  GPR[rs], clear execution and instruction hazards Place the return address link in GPR rd. The return link is the address of the second instruction following the branch,
where execution continues after a procedure call.
For processors that do not implement the MIPS16e or microMIPS ISA:
• Jump to the effective target address in GPR rs. If the target address is not 4-byte aligned, an Address Error
exception will occur when the target address is fetched.
For processors that do implement the MIPS16e or microMIPS ISA:
• Jump to the effective target address in GPR rs. Set the ISA Mode bit to the value in GPR rs bit 0. Set bit 0 of the target address to zero. If the target ISA Mode bit is 0 and the target address is not 4-byte aligned, an Address Error exception will occur when the target instruction is fetched.
In both cases, execute the instruction that follows the jump, in the branch delay slot, before executing the jump itself.
JALR.HB implements a software barrier that resolves all execution and instruction hazards created by Coprocessor 0 state changes (for Release 2 implementations, refer to the SYNCI instruction for additional information on resolving instruction hazards created by writing the instruction stream). The effects of this barrier are seen starting with the instruction fetch and decode of the instruction at the PC to which the JALR.HB instruction jumps. An equivalent bar- rier is also implemented by the ERET instruction, but that instruction is only available if access to Coprocessor 0 is enabled, whereas JALR.HB is legal in all operating modes.
This instruction clears both execution and instruction hazards. Refer to the EHB instruction description for the method of clearing execution hazards alone.
JALR.HB uses bit 10 of the instruction (the upper bit of the hint field) to denote the hazard barrier operation.
Restrictions:
JALR.HB does not clear hazards created by any instruction that is executed in the delay slot of the JALR.HB. Only hazards created by instructions executed before the JALR.HB are cleared by the JALR.HB.
After modifying an instruction stream mapping or writing to the instruction stream, execution of those instructions has UNPREDICTABLE behavior until the instruction hazard has been cleared with JALR.HB, JR.HB, ERET, or DERET. Further, the operation is UNPREDICTABLE if the mapping of the current instruction stream is modified.
SPECIAL 000000
rs
0 00000
rd
rd  00000
1
Any other legal hint value
JALR 001001
198
The MIPS32® Instruction Set Manual, Revision 6.06

JALR.HB IJump and Link Register with Hazard Barrier Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs
include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE.
Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump.
Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception.
Jump-and-Link Restartability: Register specifiers rs and rd must not be equal, because such an instruction does not have the same effect when re-executed. The result of executing such an instruction is UNPREDICTABLE. This restriction permits an exception handler to resume execution by re-executing the branch when an exception occurs in the delay slot.
Restrictions Related to Multiple Instruction Sets: This instruction can change the active instruction set, if more than one instruction set is implemented.
If only one instruction set is implemented, then the effective target address must obey the alignment rules of the instruction set. If multiple instruction sets are implemented, the effective target address must obey the alignment rules of the intended instruction set of the target address as specified by the bit 0 or GPR rs.
For processors that do not implement the microMIPS32/64 ISA, the effective target address in GPR rs must be natu- rally-aligned. For processors that do not implement the MIPS16 ASE nor microMIPS32/64 ISA, if either of the two least-significant bits are not zero, an Address Error exception occurs when the branch target is subsequently fetched as an instruction.
For processors that do implement the MIPS16 ASE or microMIPS32/64 ISA, if bit 0 is zero and bit 1 is one, an Address Error exception occurs when the jump target is subsequently fetched as an instruction.
Availability and Compatibility:
Release 6 maps JR and JR.HB to JALR and JALR.HB with rd = 0:
Pre-Release 6, JR.HB and JALR.HB were distinct instructions, both with primary opcode SPECIAL, but with distinct
function codes.
Release 6: JR.HB is defined to be JALR.HB with the destination register specifier rd set to 0. The primary opcode and function field are the same for JR.HB and JALR.HB. The pre-Release 6 instruction encoding for JR.HB is removed in Release 6.
Release 6 assemblers should accept the JR and JR.HB mnemonics, mapping them to the Release 6 instruction encod- ings.
Operation:
I: temp GPR[rs] GPR[rd]  PC + 8
I+1:if (Config3ISA = 0) and (Config1CA = 0) then PC  temp
else
PC  tempGPRLEN-1..1 || 0
ISAMode  temp0 endif
ClearHazards()
Exceptions:
None
Programming Notes:
This branch-and-link instruction can select a register for the return link; other link instructions use GPR 31. The
The MIPS32® Instruction Set Manual, Revision 6.06 199

200
The MIPS32® Instruction Set Manual, Revision 6.06
JALR.HB Jump and Link Register with Hazard Barrier default register for GPR rd, if omitted in the assembly language instruction, is GPR 31.
Release 6 JR.HB rs is implemented as JALR.HB r0,rs. For example, as JALR.HB with the destination set to the zero register, r0.
This instruction implements the final step in clearing execution and instruction hazards before execution continues. A hazard is created when a Coprocessor 0 or TLB write affects execution or the mapping of the instruction stream, or after a write to the instruction stream. When such a situation exists, software must explicitly indicate to hardware that the hazard should be cleared. Execution hazards alone can be cleared with the EHB instruction. Instruction hazards can only be cleared with a JR.HB, JALR.HB, or ERET instruction. These instructions cause hardware to clear the hazard before the instruction at the target of the jump is fetched. Note that because these instructions are encoded as jumps, the process of clearing an instruction hazard can often be included as part of a call (JALR) or return (JR) sequence, by simply replacing the original instructions with the HB equivalent.
Example: Clearing hazards due to an ASID change
/*
* Code used to modify ASID and call a routine with the new
* mapping established.
*
* a0 = New ASID to establish
* a1 = Address of the routine to call
*/
mfc0 v0, C0_EntryHi
li v1, ~M_EntryHiASID and v0, v0, v1
or v0, v0, a0
mtc0 v0, C0_EntryHi jalr.hb a1
/* Read current ASID */
/* Get negative mask for field */
/* Clear out current ASID value */
/* OR in new ASID value */
/* Rewrite EntryHi with new ASID */
/* Call routine, clearing the hazard */

JALX
31
IJump and Link Exchange 26 25 0
6 26
Format: JALX target MIPS32 with (microMIPS or MIPS16e), removed in Release 6 Purpose: Jump and Link Exchange
To execute a procedure call within the current 256 MB-aligned region and change the ISA Mode from MIPS32 to microMIPS32 or MIPS16e.
Description:
Place the return address link in GPR 31. The return link is the address of the second instruction following the branch, at which location execution continues after a procedure call. The value stored in GPR 31 bit 0 reflects the current value of the ISA Mode bit.
This is a PC-region branch (not PC-relative); the effective target address is in the “current” 256 MB-aligned region. The low 28 bits of the target address is the instr_index field shifted left 2 bits. The remaining upper bits are the corre- sponding bits of the address of the instruction in the delay slot (not the branch itself).
Jump to the effective target address, toggling the ISA Mode bit. Execute the instruction that follows the jump, in the branch delay slot, before executing the jump itself.
Restrictions:
This instruction only supports 32-bit aligned branch target addresses.
Control Transfer Instructions (CTIs) should not be placed in branch delay slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE.
Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump.
Availability and Compatibility:
If the microMIPS base architecture is not implemented and the MIPS16e ASE is not implemented, a Reserved Instruction exception is initiated.
The JALX instruction has been removed in Release 6. Pre-Release 6 code using JALX cannot run on Release 6 by trap-and-emulate. Equivalent functionality is provided by the JIALC instruction added by Release 6.
Operation:
Exceptions:
None
Programming Notes:
Forming the branch target address by concatenating PC and index bits rather than adding a signed offset to the PC is an advantage if all program code addresses fit into a 256 MB region aligned on a 256 MB boundary. It allows a branch from anywhere in the region to anywhere in the region, an action not allowed by a signed relative offset.
This definition creates the following boundary case: When the branch instruction is in the last word of a 256 MB
JALX 011101
instr_index
I: GPR[31]  PC + 8
I+1: PC  PC || instr_index || 02
GPRLEN-1..28 ISAMode  (not ISAMode)
The MIPS32® Instruction Set Manual, Revision 6.06 201

JALX IJump and Link Exchange region, it can branch only to the following 256 MB region containing the branch delay slot.
The MIPS32® Instruction Set Manual, Revision 6.06 202

JIALC IJump Indexed and Link, Compact
31
26 25 21 20 16 15 0
655 16
Format: JIALC rt, offset MIPS32 Release 6 Purpose: Jump Indexed and Link, Compact
Description: GPR[31]  PC+4, PC  ( GPR[rt] + sign_extend( offset ) )
The jump target is formed by sign extending the offset field of the instruction and adding it to the contents of GPR
rt.
The offset is NOT shifted, that is, each bit of the offset is added to the corresponding bit of the GPR.
Places the return address link in GPR 31. The return link is the address of the following instruction, where execution continues after a procedure call returns.
For processors that do not implement the MIPS16e or microMIPS ISA:
• Jump to the effective target address derived from GPR rt and the offset. If the target address is not 4-byte
aligned, an Address Error exception will occur when the target address is fetched.
For processors that do implement the MIPS16e or microMIPS ISA:
• Jump to the effective target address derived from GPR rt and the offset. Set the ISA Mode bit to bit 0 of the effec- tive address. Set bit 0 of the target address to zero. If the target ISA Mode bit is 0 and the target address is not 4- byte aligned, an Address Error exception will occur when the target instruction is fetched.
Compact jumps do not have delay slots. The instruction after the jump is NOT executed when the jump is executed.
Restrictions:
This instruction is an unconditional, always taken, compact jump, and hence has neither a delay slot nor a forbidden slot. The instruction after the jump is not executed when the jump is executed.
The register specifier may be set to the link register $31, because compact jumps do not have the restartability issues of jumps with delay slots. However, this is not common programming practice.
Availability and Compatibility:
This instruction is introduced by and required as of Release 6.
Release 6 instructions JIALC and BNEZC differ only in the rs field, instruction bits 21-25. JIALC and BNEZC
occupy the same encoding as pre-Release 6 instruction encoding SDC2, which is recoded in Release 6. Exceptions:
None
Operation:
temp  GPR[rt] + sign_extend(offset) GPR[31]  PC + 4
if (Config3ISA = 0) and (Config1CA = 0) then
PC  temp else
PC  (tempGPRLEN-1..1 || 0)
ISAMode  temp0 endif
POP76 111110
JIALC 00000
rt
offset
The MIPS32® Instruction Set Manual, Revision 6.06 203

204
The MIPS32® Instruction Set Manual, Revision 6.06
JIALC Jump Indexed and Link, Compact Programming Notes:
JIALC does NOT shift the offset before adding it the register. This can be used to eliminate tags in the least signifi- cant bits that would otherwise produce misalignment. It also allows JIALC to be used as a substitute for the JALX instruction, removed in Release 6, where the lower bits of the target PC, formed by the addition of GPR[rt] and the unshifted offset, specify the target ISAmode.

JIC
31
26 25 21 20 16 15
655 16
Format: JIC rt, offset Purpose: Jump Indexed, Compact
IJump Indexed, Compact 0
MIPS32 Release 6
POP66 110110
JIC 00000
rt
offset
Description: PC  ( GPR[rt] + sign_extend( offset ) )
The branch target is formed by sign extending the offset field of the instruction and adding it to the contents of GPR
rt.
The offset is NOT shifted, that is, each bit of the offset is added to the corresponding bit of the GPR.
For processors that do not implement the MIPS16e or microMIPS ISA:
• Jump to the effective target address derived from GPR rt and the offset. If the target address is not 4-byte aligned, an Address Error exception will occur when the target address is fetched.
For processors that do implement the MIPS16e or microMIPS ISA:
• Jump to the effective target address derived from GPR rt and the offset. Set the ISA Mode bit to bit 0 of the effec- tive address. Set bit 0 of the target address to zero. If the target ISA Mode bit is 0 and the target address is not 4- byte aligned, an Address Error exception will occur when the target instruction is fetched.
Compact jumps do not have a delay slot. The instruction after the jump is NOT executed when the jump is executed.
Restrictions:
This instruction is an unconditional, always taken, compact jump, and hence has neither a delay slot nor a forbidden slot. The instruction after the jump is not executed when the jump is executed.
Availability and Compatibility:
This instruction is introduced by and required as of Release 6.
Release 6 instructions JIC and BEQZC differ only in the rs field. JIC and BEQZC occupy the same encoding as pre- Release 6 instruction LDC2, which is recoded in Release 6.
Exceptions:
None
Operation:
temp  GPR[rt] + sign_extend(offset)
if (Config3ISA = 0) and (Config1CA = 0) then
PC  temp else
PC  (tempGPRLEN-1..1 || 0)
ISAMode  temp0 endif
Programming Notes:
JIC does NOT shift the offset before adding it the register. This can be used to eliminate tags in the least significant bits that would otherwise produce misalignment. It also allows JIALC to be used as a substitute for the JALX instruc- tion, removed in Release 6, where the lower bits of the target PC, formed by the addition of GPR[rt] and the unshifted offset, specify the target ISAmode.
The MIPS32® Instruction Set Manual, Revision 6.06 205

JR
Jump Register
pre-Release 6:
31 26 25 21 20 11 10 6 5 0
6 5 10 5 6
SPECIAL 000000
rs
0
00 0000 0000
hint
JR 001000
Release 6:
31
26 25 21 20 16 15 11 10 9 6 5 0
655556
SPECIAL 000000
rs
0 00000
00000
hint
JALR 001001
206
The MIPS32® Instruction Set Manual, Revision 6.06
Format: JR rs Purpose: Jump Register
To execute a branch to an instruction address in a register
MIPS32 Assembly idiom MIPS32 Release 6
Description: PC  GPR[rs]
Jump to the effective target address in GPR rs. Execute the instruction following the jump, in the branch delay slot,
before jumping.
For processors that do not implement the MIPS16e or microMIPS ISA:
• Jump to the effective target address in GPR rs. If the target address is not 4-byte aligned, an Address Error
exception will occur when the target address is fetched.
For processors that do implement the MIPS16e or microMIPS ISA:
• Jump to the effective target address in GPR rs. Set the ISA Mode bit to the value in GPR rs bit 0. Set bit 0 of the target address to zero. If the target ISA Mode bit is 0 and the target address is not 4-byte aligned, an Address Error exception will occur when the target instruction is fetched.
Restrictions:
Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE.
Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump.
Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception.
Restrictions Related to Multiple Instruction Sets: This instruction can change the active instruction set, if more than one instruction set is implemented.
If only one instruction set is implemented, then the effective target address must obey the alignment rules of the instruction set. If multiple instruction sets are implemented, the effective target address must obey the alignment rules of the intended instruction set of the target address as specified by the bit 0 or GPR rs.
For processors that do not implement the microMIPS ISA, the effective target address in GPR rs must be naturally- aligned. For processors that do not implement the MIPS16e ASE or microMIPS ISA, if either of the two least-signif- icant bits are not zero, an Address Error exception occurs when the branch target is subsequently fetched as an instruction.
For processors that do implement the MIPS16e ASE or microMIPS ISA, if bit 0 is zero and bit 1 is one, an Address Error exception occurs when the jump target is subsequently fetched as an instruction.

JR
IJump Register
In release 1 of the architecture, the only defined hint field value is 0, which sets default handling of JR. In Release 2 of the architecture, bit 10 of the hint field is used to encode an instruction hazard barrier. See the JR.HB instruction description for additional information.
Availability and Compatibility:
Release 6 maps JR and JR.HB to JALR and JALR.HB with rd = 0:
Pre-Release 6, JR and JALR were distinct instructions, both with primary opcode SPECIAL, but with distinct func- tion codes.
Release 6: JR is defined to be JALR with the destination register specifier rd set to 0. The primary opcode and func- tion field are the same for JR and JALR. The pre-Release 6 instruction encoding for JR is removed in Release 6.
Release 6 assemblers should accept the JR and JR.HB mnemonics, mapping them to the Release 6 instruction encod- ings.
Operation:
I: temp  GPR[rs]
I+1:if (Config3ISA = 0) and (Config1CA = 0) then
PC  temp else
PC  tempGPRLEN-1..1 || 0 ISAMode  temp0
endif
Exceptions:
None
Programming Notes:
Software should use the value 31 for the rs field of the instruction word on return from a JAL, JALR, or BGEZAL, and should use a value other than 31 for remaining uses of JR.
The MIPS32® Instruction Set Manual, Revision 6.06 207

JR.HB Jump Register with Hazard Barrier pre-Release 6:
31 26 25 21 20 11 10 9 6 5 0
6 5 10 14 6
SPECIAL 000000
rs
0
00 0000 0000
1
Any other legal hint value
JR 001000
Release 6:
31
26 25 21 20 16 15 11 10 9 6 5 0
6555146
Format: JR.HB rs MIPS32 Release 2 Assembly idiom Release 6
Purpose: Jump Register with Hazard Barrier
To execute a branch to an instruction address in a register and clear all execution and instruction hazards.
Description: PC  GPR[rs], clear execution and instruction hazards
Jump to the effective target address in GPR rs. Execute the instruction following the jump, in the branch delay slot,
before jumping.
For processors that do not implement the MIPS16e or microMIPS ISA:
• Jump to the effective target address in GPR rs. If the target address is not 4-byte aligned, an Address Error
exception will occur when the target address is fetched.
For processors that do implement the MIPS16e or microMIPS ISA:
• Jump to the effective target address in GPR rs. Set the ISA Mode bit to the value in GPR rs bit 0. Set bit 0 of the target address to zero. If the target ISA Mode bit is 0 and the target address is not 4-byte aligned, an Address Error exception will occur when the target instruction is fetched.
JR.HB implements a software barrier that resolves all execution and instruction hazards created by Coprocessor 0 state changes (for Release 2 implementations, refer to the SYNCI instruction for additional information on resolving instruction hazards created by writing the instruction stream). The effects of this barrier are seen starting with the instruction fetch and decode of the instruction at the PC to which the JR.HB instruction jumps. An equivalent barrier is also implemented by the ERET instruction, but that instruction is only available if access to Coprocessor 0 is enabled, whereas JR.HB is legal in all operating modes.
This instruction clears both execution and instruction hazards. Refer to the EHB instruction description for the method of clearing execution hazards alone.
JR.HB uses bit 10 of the instruction (the upper bit of the hint field) to denote the hazard barrier operation.
Restrictions:
JR.HB does not clear hazards created by any instruction that is executed in the delay slot of the JR.HB. Only hazards created by instructions executed before the JR.HB are cleared by the JR.HB.
After modifying an instruction stream mapping or writing to the instruction stream, execution of those instructions has UNPREDICTABLE behavior until the hazard has been cleared with JALR.HB, JR.HB, ERET, or DERET. Fur- ther, the operation is UNPREDICTABLE if the mapping of the current instruction stream is modified.
Control Transfer Instructions (CTIs) should not be placed in branch delay slots or Release 6 forbidden slots. CTIs The MIPS32® Instruction Set Manual, Revision 6.06
SPECIAL 000000
rs
0 00000
0 00000
1
Any other legal hint value
JALR 001001
208

JR.HB IJump Register with Hazard Barrier include all branches and jumps, NAL, ERET, ERETNC, DERET, WAIT, and PAUSE.
Pre-Release 6: Processor operation is UNPREDICTABLE if a control transfer instruction (CTI) is placed in the delay slot of a branch or jump.
Release 6: If a control transfer instruction (CTI) is executed in the delay slot of a branch or jump, Release 6 imple- mentations are required to signal a Reserved Instruction exception.
Restrictions Related to Multiple Instruction Sets: This instruction can change the active instruction set, if more than one instruction set is implemented.
If only one instruction set is implemented, then the effective target address must obey the alignment rules of the instruction set. If multiple instruction sets are implemented, the effective target address must obey the alignment rules of the intended instruction set of the target address as specified by the bit 0 or GPR rs.
For processors that do not implement the microMIPS ISA, the effective target address in GPR rs must be naturally- aligned. For processors that do not implement the MIPS16 ASE or microMIPS ISA, if either of the two least-signifi- cant bits are not zero, an Address Error exception occurs when the branch target is subsequently fetched as an instruc- tion.
For processors that do implement the MIPS16 ASE or microMIPS ISA, if bit 0 is zero and bit 1 is one, an Address Error exception occurs when the jump target is subsequently fetched as an instruction.
Availability and Compatibility:
Release 6 maps JR and JR.HB to JALR and JALR.HB with rd = 0:
Pre-Release 6, JR.HB and JALR.HB were distinct instructions, both with primary opcode SPECIAL, but with distinct function codes.
Release 6: JR.HB is defined to be JALR.HB with the destination register specifier rd set to 0. The primary opcode and function field are the same for JR.HB and JALR.HB. The pre-Release 6 instruction encoding for JR.HB is removed in Release 6.
Release 6 assemblers should accept the JR and JR.HB mnemonics, mapping them to the Release 6 instruction encod- ings.
Operation:
I: temp  GPR[rs]
I+1:if (Config3ISA = 0) and (Config1CA = 0) then
PC  temp else
PC  tempGPRLEN-1..1 || 0 ISAMode  temp0
endif
ClearHazards()
Exceptions:
None
Programming Notes:
This instruction implements the final step in clearing execution and instruction hazards before execution continues. A hazard is created when a Coprocessor 0 or TLB write affects execution or the mapping of the instruction stream, or after a write to the instruction stream. When such a situation exists, software must explicitly indicate to hardware that the hazard should be cleared. Execution hazards alone can be cleared with the EHB instruction. Instruction hazards can only be cleared with a JR.HB, JALR.HB, or ERET instruction. These instructions cause hardware to clear the hazard before the instruction at the target of the jump is fetched. Note that because these instructions are encoded as jumps, the process of clearing an instruction hazard can often be included as part of a call (JALR) or return (JR)
The MIPS32® Instruction Set Manual, Revision 6.06 209

210
The MIPS32® Instruction Set Manual, Revision 6.06
JR.HB Jump Register with Hazard Barrier sequence, by simply replacing the original instructions with the HB equivalent.
Example: Clearing hazards due to an ASID change
/*
* Routine called to modify ASID and return with the new * mapping established.
*
* a0 = New ASID to establish
*/
mfc0 v0, C0_EntryHi
li v1, ~M_EntryHiASID and v0, v0, v1
or v0, v0, a0
mtc0 v0, C0_EntryHi jr.hb ra
nop
/* Read current ASID */
/* Get negative mask for field */
/* Clear out current ASID value */
/* OR in new ASID value */
/* Rewrite EntryHi with new ASID */
/* Return, clearing the hazard */
Example: Making a write to the instruction stream visible
/*
* Routine called after new instructions are written to
* make them visible and return with the hazards cleared.
*/
{Synchronize the caches – see the SYNCI and CACHE instructions}
sync /* jr.hb ra /* nop
Example: Clearing instruction hazards in-line
la AT, 10f
jr.hb AT /* nop /*
Force memory synchronization */
Return, clearing the hazard */
Jump to next instruction, clearing */
hazards */
10:

LB
31
26 25 21 20 16 15
655 16
Format: LB rt, offset(base) Purpose: Load Byte
To load a byte from memory as a signed value.
ILoad Byte 0
MIPS32
LB 100000
base
rt
offset
Description: GPR[rt]  memory[GPR[base] + offset]
The contents of the 8-bit byte at the memory location specified by the effective address are fetched, sign-extended,
and placed in GPR rt. The 16-bit signed offset is added to the contents of GPR base to form the effective address. Restrictions:
None
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD)
pAddr  pAddr || (pAddr xor ReverseEndian2) PSIZE-1..2 1..0
memword  LoadMemory (CCA, BYTE, pAddr, vAddr, DATA)
byte  vAddr Exceptions:
xor BigEndianCPU2
GPR[rt]  sign_extend(memword7+8*byte..8*byte)
1..0
TLB Refill, TLB Invalid, Address Error, Watch
The MIPS32® Instruction Set Manual, Revision 6.06 211

LBE
Load Byte EVA
31
26 25 21 20 16 15 7 6 5 0
655916
Format: LBE rt, offset(base) MIPS32 Purpose: Load Byte EVA
To load a byte as a signed value from user mode virtual address space when executing in kernel mode.
Description: GPR[rt]  memory[GPR[base] + offset]
The contents of the 8-bit byte at the memory location specified by the effective address are fetched, sign-extended,
and placed in GPR rt. The 9-bit signed offset is added to the contents of GPR base to form the effective address.
The LBE instruction functions the same as the LB instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment config- ured to use the MUSUK access mode and executing in kernel mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Enhanced Virtual Addressing section for additional informa- tion.
Implementation of this instruction is specified by the Config5EVA field being set to one. Restrictions:
Only usable when access to Coprocessor0 is enabled and accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD)
pAddr  pAddr || (pAddr xor ReverseEndian2) PSIZE-1..2 1..0
memword  LoadMemory (CCA, BYTE, pAddr, vAddr, DATA)
Exceptions:
TLB Refill, TLB Invalid
Bus Error, Address Error, Watch, Reserved Instruction, Coprocessor Unusable
SPECIAL3 011111
base
rt
offset
0
LBE 101100
212
The MIPS32® Instruction Set Manual, Revision 6.06
byte  vAddr
xor BigEndianCPU2
GPR[rt]  sign_extend(memword7+8*byte..8*byte)
1..0

LBU
31
26 25 21 20 16 15
655 16
Format: LBU rt, offset(base) Purpose: Load Byte Unsigned
To load a byte from memory as an unsigned value
ILoad Byte Unsigned 0
MIPS32
LBU 100100
base
rt
offset
Description: GPR[rt]  memory[GPR[base] + offset]
The contents of the 8-bit byte at the memory location specified by the effective address are fetched, zero-extended,
and placed in GPR rt. The 16-bit signed offset is added to the contents of GPR base to form the effective address. Restrictions:
None
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD)
pAddr  pAddr || (pAddr xor ReverseEndian2) PSIZE-1..2 1..0
memword  LoadMemory (CCA, BYTE, pAddr, vAddr, DATA)
byte  vAddr Exceptions:
xor BigEndianCPU2
GPR[rt]  zero_extend(memword7+8*byte..8*byte)
1..0
TLB Refill, TLB Invalid, Address Error, Watch
The MIPS32® Instruction Set Manual, Revision 6.06 213

LBUE Load Byte Unsigned EVA
31
26 25 21 20 16 15 7 6 5 0
655916
Format: LBUE rt, offset(base) MIPS32 Purpose: Load Byte Unsigned EVA
To load a byte as an unsigned value from user mode virtual address space when executing in kernel mode.
Description: GPR[rt]  memory[GPR[base] + offset]
The contents of the 8-bit byte at the memory location specified by the effective address are fetched, zero-extended,
and placed in GPR rt. The 9-bit signed offset is added to the contents of GPR base to form the effective address.
The LBUE instruction functions the same as the LBU instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment con- figured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Enhanced Virtual Addressing section for additional information.
Implementation of this instruction is specified by the Config5EVA field being set to one. Restrictions:
Only usable when access to Coprocessor0 is enabled and accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD)
pAddr  pAddr || (pAddr xor ReverseEndian2) PSIZE-1..2 1..0
memword  LoadMemory (CCA, BYTE, pAddr, vAddr, DATA)
SPECIAL3 011111
base
rt
offset
0
LBUE 101000
214
The MIPS32® Instruction Set Manual, Revision 6.06
byte  vAddr Exceptions:
xor BigEndianCPU2
GPR[rt]  zero_extend(memword7+8*byte..8*byte)
1..0
TLB Refill, TLB Invalid, Bus Error, Address Error, Watch, Reserved Instruction, Coprocessor Unusable

LDC1
ILoad Doubleword to Floating Point
31
26 25 21 20 16 15 0
655 16
Format: LDC1 ft, offset(base) MIPS32 Purpose: Load Doubleword to Floating Point
To load a doubleword from memory to an FPR.
Description: FPR[ft]  memory[GPR[base] + offset]
The contents of the 64-bit doubleword at the memory location specified by the aligned effective address are fetched
and placed in FPR ft. The 16-bit signed offset is added to the contents of GPR base to form the effective address.
Restrictions:
Pre-Release 6: An Address Error exception occurs if EffectiveAddress2..0 ≠ 0 (not doubleword-aligned). Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Operation:
vAddr sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) paddr  paddr xor ((BigEndianCPU xor ReverseEndian) || 02) memlsw  LoadMemory(CCA, WORD, pAddr, vAddr, DATA)
paddr  paddr xor 0b100
memmsw  LoadMemory(CCA, WORD, pAddr, vAddr+4, DATA) memdoubleword  memmsw || memlsw
StoreFPR(ft, UNINTERPRETED_DOUBLEWORD, memdoubleword)
Exceptions:
Coprocessor Unusable, Reserved Instruction, TLB Refill, TLB Invalid, Address Error, Watch
LDC1 110101
base
ft
offset
The MIPS32® Instruction Set Manual, Revision 6.06 215

LDC2 Load Doubleword to Coprocessor 2 pre-Release 6
31 26 25 21 20 16 15 0
655 16
LDC2 110110
base
rt
offset
Release 6
31
26 25 21 20 16 15 11 10 0
COP2 010010
LDC2 01110
rt
base
offset
216
The MIPS32® Instruction Set Manual, Revision 6.06
6 5 5 5 11
Format: LDC2 rt, offset(base) Purpose: Load Doubleword to Coprocessor 2
To load a doubleword from memory to a Coprocessor 2 register.
Description: CPR[2,rt,0]  memory[GPR[base] + offset]
MIPS32
The contents of the 64-bit doubleword at the memory location specified by the aligned effective address are fetched and placed in Coprocessor 2 register rt. The signed offset is added to the contents of GPR base to form the effective address.
Restrictions:
Pre-Release 6: An Address Error exception occurs if EffectiveAddress2..0 ≠ 0 (not doubleword-aligned). Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Availability and Compatibility:
This instruction has been recoded for Release 6.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) paddr  paddr xor ((BigEndianCPU xor ReverseEndian) || 02) memlsw  LoadMemory(CCA, WORD, pAddr, vAddr, DATA)
paddr  paddr xor 0b100
memmsw  LoadMemory(CCA, WORD, pAddr, vAddr+4, DATA) memlsw
memmsw
Exceptions:
Coprocessor Unusable, Reserved Instruction, TLB Refill, TLB Invalid, Address Error, Watch
Programming Notes:
Release 6 implements a 9-bit offset, whereas all release levels lower than Release 6 implement a 16-bit offset.
Programming Notes:
As shown in the instruction drawing above, Release 6 implements an 11-bit offset, whereas all release levels lower

LDC2 ILoad Doubleword to Coprocessor 2 than Release 6 of the MIPS architecture implement a 16-bit offset.
The MIPS32® Instruction Set Manual, Revision 6.06 217

LDXC1 Load Doubleword Indexed to Floating Point
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: LDXC1 fd, index(base) MIPS32 Release 2 removed in Release 6 Purpose: Load Doubleword Indexed to Floating Point
To load a doubleword from memory to an FPR (GPR+GPR addressing)
Description: FPR[fd]  memory[GPR[base] + GPR[index]]
The contents of the 64-bit doubleword at the memory location specified by the aligned effective address are fetched
and placed in FPR fd. The contents of GPR index and GPR base are added to form the effective address. Restrictions:
An Address Error exception occurs if EffectiveAddress2..0 ≠ 0 (not doubleword-aligned).
Availability and Compatibility:
This instruction has been removed in Release 6.
Required in all versions of MIPS64 since MIPS64 Release 1. Not available in MIPS32 Release 1. Required in MIPS32 Release 2 and all subsequent versions of MIPS32. When required, required whenever FPU is present, whether a 32-bit or 64-bit FPU, whether in 32-bit or 64-bit FP Register Mode (FIRF64=0 or 1, StatusFR=0 or 1).
Operation:
COP1X 010011
base
index
0 00000
fd
LDXC1 000001
218
The MIPS32® Instruction Set Manual, Revision 6.06
vAddr GPR[base] + GPR[index] if vAddr  03 then
2..0 SignalException(AddressError)
endif
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) paddr  paddr xor ((BigEndianCPU xor ReverseEndian) || 02) memlsw  LoadMemory(CCA, WORD, pAddr, vAddr, DATA)
paddr  paddr xor 0b100
memmsw  LoadMemory(CCA, WORD, pAddr, vAddr+4, DATA) memdoubleword  memmsw || memlsw
StoreFPR(fd, UNINTERPRETED_DOUBLEWORD, memdoubleword)
Exceptions:
TLB Refill, TLB Invalid, Address Error, Reserved Instruction, Coprocessor Unusable, Watch

LH
31
26 25 21 20 16 15
655 16
Format: LH rt, offset(base) Purpose: Load Halfword
To load a halfword from memory as a signed value
Description: GPR[rt]  memory[GPR[base] + offset]
ILoad Halfword 0
MIPS32
LH 100001
base
rt
offset
The contents of the 16-bit halfword at the memory location specified by the aligned effective address are fetched, sign-extended, and placed in GPR rt. The 16-bit signed offset is added to the contents of GPR base to form the effec- tive address.
Restrictions:
Pre-Release 6: The effective address must be naturally-aligned. If the least-significant bit of the address is non-zero, an Address Error exception occurs.
Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD)
pAddr  pAddrPSIZE-1..2 || (pAddr1..0 xor (ReverseEndian || 0)) memword  LoadMemory (CCA, HALFWORD, pAddr, vAddr, DATA) byte  vAddr1..0 xor (BigEndianCPU || 0)
GPR[rt]  sign_extend(memword15+8*byte..8*byte)
Exceptions:
TLB Refill, TLB Invalid, Bus Error, Address Error, Watch
The MIPS32® Instruction Set Manual, Revision 6.06 219

LHE
ILoad Halfword EVA
31
26 25 21 20 16 15 7 6 5 0
655916
Format: LHE rt, offset(base) MIPS32 Purpose: Load Halfword EVA
To load a halfword as a signed value from user mode virtual address space when executing in kernel mode.
Description: GPR[rt]  memory[GPR[base] + offset]
The contents of the 16-bit halfword at the memory location specified by the aligned effective address are fetched, sign-extended, and placed in GPR rt. The 9-bit signed offset is added to the contents of GPR base to form the effec- tive address.
The LHE instruction functions the same as the LH instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment config- ured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Enhanced Virtual Addressing section for additional information.
Implementation of this instruction is specified by the Config5EVA field being set to one. Restrictions:
Only usable when access to Coprocessor0 is enabled and accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode.
Pre-Release 6: The effective address must be naturally-aligned. If the least-significant bit of the address is non-zero, an Address Error exception occurs.
Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD)
pAddr  pAddrPSIZE-1..2 || (pAddr1..0 xor (ReverseEndian || 0)) memword  LoadMemory (CCA, HALFWORD, pAddr, vAddr, DATA) byte  vAddr1..0 xor (BigEndianCPU || 0)
GPR[rt]  sign_extend(memword15+8*byte..8*byte)
Exceptions:
TLB Refill, TLB Invalid, Bus Error, Address Error Watch, Reserved Instruction, Coprocessor Unusable
SPECIAL3 011111
base
rt
offset
0
LHE 101101
The MIPS32® Instruction Set Manual, Revision 6.06 220

LHU
31
26 25 21 20 16 15
655 16
Format: LHU rt, offset(base) Purpose: Load Halfword Unsigned
To load a halfword from memory as an unsigned value
Description: GPR[rt]  memory[GPR[base] + offset]
ILoad Halfword Unsigned 0
MIPS32
LHU 100101
base
rt
offset
The contents of the 16-bit halfword at the memory location specified by the aligned effective address are fetched, zero-extended, and placed in GPR rt. The 16-bit signed offset is added to the contents of GPR base to form the effec- tive address.
Restrictions:
Pre-Release 6: The effective address must be naturally-aligned. If the least-significant bit of the address is non-zero, an Address Error exception occurs.
Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD)
pAddr  pAddrPSIZE-1..2 || (pAddr1..0 xor (ReverseEndian || 0)) memword  LoadMemory (CCA, HALFWORD, pAddr, vAddr, DATA) byte  vAddr1..0 xor (BigEndianCPU || 0)
GPR[rt]  zero_extend(memword15+8*byte..8*byte)
Exceptions:
TLB Refill, TLB Invalid, Address Error, Watch
The MIPS32® Instruction Set Manual, Revision 6.06 221

LHUE Load Halfword Unsigned EVA
31
26 25 21 20 16 15 7 6 5 0
655916
Format: LHUE rt, offset(base) MIPS32 Purpose: Load Halfword Unsigned EVA
To load a halfword as an unsigned value from user mode virtual address space when executing in kernel mode.
Description: GPR[rt]  memory[GPR[base] + offset]
The contents of the 16-bit halfword at the memory location specified by the aligned effective address are fetched, zero-extended, and placed in GPR rt. The 9-bit signed offset is added to the contents of GPR base to form the effec- tive address.
The LHUE instruction functions the same as the LHU instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment con- figured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Enhanced Virtual Addressing section for additional information.
Implementation of this instruction is specified by the Config5EVA field being set to one. Restrictions:
Only usable when access to Coprocessor0 is enabled and accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode.
Pre-Release 6: The effective address must be naturally-aligned. If the least-significant bit of the address is non-zero, an Address Error exception occurs.
Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD)
pAddr  pAddrPSIZE-1..2 || (pAddr1..0 xor (ReverseEndian || 0)) memword  LoadMemory (CCA, HALFWORD, pAddr, vAddr, DATA) byte  vAddr1..0 xor (BigEndianCPU || 0)
GPR[rt]  zero_extend(memword15+8*byte..8*byte)
Exceptions:
TLB Refill, TLB Invalid, Bus Error, Address Error, Watch, Reserved Instruction, Coprocessor Unusable
SPECIAL3 011111
base
rt
offset
0
LHUE 101001
222
The MIPS32® Instruction Set Manual, Revision 6.06

LL
ILoad Linked Word
pre-Release 6
31
Release 6
26 25 21 20 16 15
655 16
0
7 6 5 0
16
MIPS32
Description: GPR[rt]  memory[GPR[base] + offset]
The LL and SC instructions provide the primitives to implement atomic read-modify-write (RMW) operations for
synchronizable memory locations.
The contents of the 32-bit word at the memory location specified by the aligned effective address are fetched and
written into GPR rt. The signed offset is added to the contents of GPR base to form an effective address.
This begins a RMW sequence on the current processor. There can be only one active RMW sequence per processor. When an LL is executed it starts an active RMW sequence replacing any other sequence that was active. The RMW sequence is completed by a subsequent SC instruction that either completes the RMW sequence atomically and suc- ceeds, or does not and fails.
Executing LL on one processor does not cause an action that, by itself, causes an SC for the same block to fail on another processor.
An execution of LL does not have to be followed by execution of SC; a program is free to abandon the RMW sequence without attempting a write.
Restrictions:
The addressed location must be synchronizable by all processors and I/O devices sharing the location; if it is not, the result is UNPREDICTABLE. Which storage is synchronizable is a function of both CPU and system implementa- tions. See the documentation of the SC instruction for the formal definition.
The effective address must be naturally-aligned. If either of the 2 least-significant bits of the effective address is non- zero, an Address Error exception occurs.
Providing misaligned support for Release 6 is not a requirement for this instruction.
Availability and Compatibility:
This instruction has been reallocated an opcode in Release 6.
Operation:
vAddr  sign_extend(offset) + GPR[base] if vAddr1..0 ≠ 02 then
SignalException(AddressError)
endif
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD)
The MIPS32® Instruction Set Manual, Revision 6.06 223
LL 110000
base
rt
offset
31
26 25 21 20 16 15
6559
Format: LL rt, offset(base) Purpose: Load Linked Word
To load a word from memory for an atomic read-modify-write
SPECIAL3 011111
base
rt
offset
0
LL 110110

LL
ILoad Linked Word
memword  LoadMemory (CCA, WORD, pAddr, vAddr, DATA) GPR[rt] memword
LLbit  1
Exceptions:
TLB Refill, TLB Invalid, Address Error, Watch
Programming Notes:
Release 6 implements a 9-bit offset, whereas all release levels lower than Release 6 implement a 16-bit offset.
The MIPS32® Instruction Set Manual, Revision 6.06 224

LLE
Load Linked Word EVA
31
26 25 21 20 16 15 7 6 5 0
655916
Format: LLE rt, offset(base) MIPS32 Purpose: Load Linked Word EVA
To load a word from a user mode virtual address when executing in kernel mode for an atomic read-modify-write
Description: GPR[rt]  memory[GPR[base] + offset]
The LLE and SCE instructions provide the primitives to implement atomic read-modify-write (RMW) operations for
synchronizable memory locations using user mode virtual addresses while executing in kernel mode.
The contents of the 32-bit word at the memory location specified by the aligned effective address are fetched and
written into GPR rt. The 9-bit signed offset is added to the contents of GPR base to form an effective address.
This begins a RMW sequence on the current processor. There can be only one active RMW sequence per processor. When an LLE is executed it starts an active RMW sequence replacing any other sequence that was active. The RMW sequence is completed by a subsequent SCE instruction that either completes the RMW sequence atomically and suc- ceeds, or does not and fails.
Executing LLE on one processor does not cause an action that, by itself, causes an SCE for the same block to fail on another processor.
An execution of LLE does not have to be followed by execution of SCE; a program is free to abandon the RMW sequence without attempting a write.
The LLE instruction functions the same as the LL instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment config- ured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Segmentation Control for additional information.
Implementation of this instruction is specified by the Config5EVA field being set to one. Restrictions:
The addressed location must be synchronizable by all processors and I/O devices sharing the location; if it is not, the result is UNPREDICTABLE. Which storage is synchronizable is a function of both CPU and system implementa- tions. See the documentation of the SCE instruction for the formal definition.
The effective address must be naturally-aligned. If either of the 2 least-significant bits of the effective address is non- zero, an Address Error exception occurs.
Providing misaligned support for Release 6 is not a requirement for this instruction.
Operation:
vAddr  sign_extend(offset) + GPR[base] if vAddr1..0 ≠ 02 then
SignalException(AddressError)
endif
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) memword  LoadMemory (CCA, WORD, pAddr, vAddr, DATA) GPR[rt]  memword
LLbit  1
SPECIAL3 011111
base
rt
offset
0
LLE 101110
225
The MIPS32® Instruction Set Manual, Revision 6.06

LLE
ILoad Linked Word EVA TLB Refill, TLB Invalid, Address Error, Reserved Instruction, Watch, Coprocessor Unusable
Exceptions: Programming Notes:
The MIPS32® Instruction Set Manual, Revision 6.06 226

LLE Load Linked Word EVA
227 The MIPS32® Instruction Set Manual, Revision 6.06

LLWP ILoad Linked Word Paired
31
26 25 21 20 16 15 11 10 7 6 5 0
6555416
Format: LLWP rt, rd, (base) MIPS32 Release 6 Purpose: Load Linked Word Paired
To load two words from memory for an atomic read-modify-write, writing a word each to two registers.
Description: GPR[rd]  memory[GPR[base]]63..32, GPR[rt]  memory[GPR[base]]31..0
The LLWP and SCWP instructions provide primitives to implement a paired word atomic read-modify-write (RMW)
operation at a synchronizable memory location.
The 64-bit paired word, as a concatenation of two words, at the memory location specified by the double-word aligned effective address is read. The least significant word is written into GPR rt,and the most significant word is written into GPR rd.
A paired word read or write occurs as a pair of word reads or writes that is double-word atomic.
The instruction has no offset. The effective address is equal to the contents of GPR base.
The execution of LLWP begins a RMW sequence on the current processor. There can be only one active RMW sequence per processor. When an LLWP is executed it starts an active RMW sequence replacing any other sequence that was active. The RMW sequence is completed by a subsequent SCWP instruction that either completes the RMW sequence atomically and succeeds, or does not and fails.
Successful execution of the LLWP results in setting LLbit and writing CP0 LLAddr, where LLbit is the least-signif- icant bit of LLAddr. LLAddr contains the data-type aligned address of the operation, in this case a double-word.
Executing LLWP on one processor does not cause an action that, by itself, causes a store conditional instruction type for the same block to fail on another processor.
An execution of LLWP does not have to be followed by execution of SCWP; a program is free to abandon the RMW sequence without attempting a write.
Restrictions:
The addressed location must be synchronizable by all processors and I/O devices sharing the location; if it is not, the result is UNPREDICTABLE. Which storage is synchronizable is a function of both CPU and system implementa- tions. See the documentation of the SC instruction for the formal definition.
The architecture optionally allows support for Load-Linked and Store-Conditional instruction types in a cacheless processor. Support for cacheless operation is implementation dependent. In this case, LLAddr is optional.
Providing misaligned support is not a requirement for this instruction.
Availability and Compatibility
This instruction is introduced by Release 6. It is only present if Config5XNP=0. Operation:
vAddr  GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD)
// PAIREDWORD: two word data-type that is double-word atomic memdoubleword  LoadMemory (CCA, PAIREDWORD, pAddr, vAddr, DATA)
SPECIAL3 011111
base
rt
rd
0 0000
1
LL 110110
The MIPS32® Instruction Set Manual, Revision 6.06 228

229
The MIPS32® Instruction Set Manual, Revision 6.06
LLWP
Load Linked Word Paired
GPR[rt]  memdoubleword31..0
GPR[rd]  memdoubleword63..32
LLAddr  pAddr // double-word aligned i.e., pAddr2..0 are 0, or not supported. LLbit  1
Exceptions:
TLB Refill, TLB Invalid, Reserved Instruction, Address Error, Watch
Programming Notes:
An LLWP instruction for which the two destination registers are the same but non-zero is UNPREDICTABLE. An LLWP with two zero destination registers followed by a SCWP can be used to accomplish a double-word atomic write.

LLWPE ILoad Linked Word Paired EVA
31
26 25 21 20 16 15 11 10 7 6 5 0
6555416
Format: LLWPE rt, rd, (base) MIPS32 Release 6 Purpose: Load Linked Word Paired EVA
To load two words from memory for an atomic read-modify-write, writing a word each to two registers. The load occurs in kernel mode from user virtual address space.
Description: GPR[rd]  memory[GPR[base]]63..32, GPR[rt]  memory[GPR[base]]31..0
The LLWPE and SCWPE instructions provide primitives to implement a paired word atomic read-modify-write
(RMW) operation at a synchronizable memory location.
The 64-bit paired word at the memory location specified by the double-word aligned effective address is read. The
least significant word is written into GPR rt. The most significant word is written into GPR rd. A paired word read or write occurs as a pair of word reads or writes that is double-word atomic. The instruction has no offset. The effective address is equal to the contents of GPR base.
The execution of LLWPE begins a RMW sequence on the current processor. There can be only one active RMW sequence per processor. When an LLWPE is executed it starts an active RMW sequence replacing any other sequence that was active. The RMW sequence is completed by a subsequent SCWPE instruction that either completes the RMW sequence atomically and succeeds, or does not and fails.
Successful execution of the LLWPE results in setting LLbit and writing CP0 LLAddr, where LLbit is the least-sig- nificant bit of LLAddr. LLAddr contains the data-type aligned address of the operation, in this case a double-word aligned address.
The LLWPE instruction functions the same as the LLWP instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment configured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also acces- sible. Refer to Volume III, Segmentation Control for additional information.
Executing LLWPE on one processor does not cause an action that, by itself, causes a store conditional instruction type for the same block to fail on another processor.
An execution of LLWPE does not have to be followed by execution of SCWPE; a program is free to abandon the RMW sequence without attempting a write.
Restrictions:
The addressed location must be synchronizable by all processors and I/O devices sharing the location; if it is not, the result is UNPREDICTABLE. Which storage is synchronizable is a function of both CPU and system implementa- tions. See the documentation of the SC instruction for the formal definition.
The architecture optionally allows support for Load-Linked and Store-Conditional instruction types in a cacheless processor. Support for cacheless operation is implementation dependent. In this case, LLAddr is optional.
Providing misaligned support is not a requirement for this instruction.
Availability and Compatibility
This instruction is introduced by Release 6. It is only present if Config5XNP=0 and Config5EVA=1.
SPECIAL3 011111
base
rt
rd
0 0000
1
LLE 101110
The MIPS32® Instruction Set Manual, Revision 6.06 230

231
The MIPS32® Instruction Set Manual, Revision 6.06
LLWPE Load Linked Word Paired EVA Operation:
vAddr  GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD)
// PAIREDWORD: two word data-type that is double-word atomic
memdoubleword  LoadMemory (CCA, PAIREDWORD, pAddr, vAddr, DATA)
GPR[rt]  memdoubleword31..0
GPR[rd]  memdoubleword63..32
LLAddr  pAddr // double-word aligned i.e., pAddr2..0 are 0, or not supported. LLbit  1
Exceptions:
TLB Refill, TLB Invalid, Reserved Instruction, Address Error, Watch, Coprocessor Unusable.
Programming Notes:
An LLWPE instruction for which the two destination registers are the same but non-zero is UNPREDICTABLE. An LLWPE with two zero destination registers followed by a SCWPE can be used to accomplish a double-word atomic write.

LSA
ILoad Scaled Address
31
26 25 21 20 16 15 11 10 8 7 6 5 0
6555326
Format: LSA
LSA rd,rs,rt,sa MIPS32 Release 6
Purpose: Load Scaled Address
Description:
GPR[rd]  sign_extend.32( (GPR[rs] << (sa+1)) + GPR[rt] ) LSA adds two values derived from registers rs and rt, with a scaling shift on rs. The scaling shift is formed by adding 1 to the 2-bit sa field, which is interpreted as unsigned. The scaling left shift varies from 1 to 5, corresponding to multiplicative scaling values of 2, 4, 8, 16, bytes, or 16, 32, 64, or 128 bits. Restrictions: None Availability and Compatibility: LSA instruction is introduced by and required as of Release 6. Operation GPR[rd]  sign_extend.32( GPR[rs] << (sa+1) + GPR[rt] ) Exceptions: None SPECIAL 000000 rs rt rd 000 sa LSA 000101 The MIPS32® Instruction Set Manual, Revision 6.06 232 LUI Load Upper Immediate Pre-Release 6 31 26 25 21 20 16 15 0 0 LUI 001111 0 00000 rt immediate Release 6 31 26 25 21 20 16 15 655 16 655 16 AUI 001111 00000 rt immediate 233 The MIPS32® Instruction Set Manual, Revision 6.06 Format: LUI rt, immediate MIPS32, Assembly Idiom Release 6 Purpose: Load Upper Immediate To load a constant into the upper half of a word Description: GPR[rt]  immediate || 016 The 16-bit immediate is shifted left 16 bits and concatenated with 16 bits of low-order zeros. The 32-bit result is placed into GPR rt. Restrictions: None. Operation: GPR[rt]  immediate || 016 Exceptions: None Programming Notes: In Release 6, LUI is an assembly idiom of AUI with rs=0. LUXC1 ILoad Doubleword Indexed Unaligned to Floating Point 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: LUXC1 fd, index(base) MIPS32 Release 2, removed in Release 6 Purpose: Load Doubleword Indexed Unaligned to Floating Point To load a doubleword from memory to an FPR (GPR+GPR addressing), ignoring alignment Description: FPR[fd]  memory[(GPR[base] + GPR[index])PSIZE-1..3] The contents of the 64-bit doubleword at the memory location specified by the effective address are fetched and placed into the low word of FPR fd. The contents of GPR index and GPR base are added to form the effective address. The effective address is doubleword-aligned; EffectiveAddress2..0 are ignored. Restrictions: The result of this instruction is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model; it is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: This instruction has been removed in Release 6. Operation: COP1X 010011 base index 0 00000 fd LUXC1 000101 vAddr  (GPR[base]+GPR[index]) (pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) paddr  paddr xor ((BigEndianCPU xor ReverseEndian) || 02) memlsw  LoadMemory(CCA, WORD, pAddr, vAddr, DATA) paddr  paddr xor 0b100 memmsw  LoadMemory(CCA, WORD, pAddr, vAddr+4, DATA) memdoubleword  memmsw || memlsw StoreFPR(ft, UNINTERPRETED_DOUBLEWORD, memdoubleword) Exceptions: Coprocessor Unusable, Reserved Instruction, TLB Refill, TLB Invalid, Watch 31..3 || 03 The MIPS32® Instruction Set Manual, Revision 6.06 234 LW Load Word 31 26 25 21 20 16 15 655 16 Format: LW rt, offset(base) Purpose: Load Word To load a word from memory as a signed value Description: GPR[rt]  memory[GPR[base] + offset] 0 MIPS32 LW 100011 base rt offset 235 The MIPS32® Instruction Set Manual, Revision 6.06 The contents of the 32-bit word at the memory location specified by the aligned effective address are fetched, sign- extended to the GPR register length if necessary, and placed in GPR rt. The 16-bit signed offset is added to the con- tents of GPR base to form the effective address. Restrictions: Pre-Release 6: The effective address must be naturally-aligned. If either of the 2 least-significant bits of the address is non-zero, an Address Error exception occurs. Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment. Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent. Operation: vAddr  sign_extend(offset) + GPR[base] (pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) memword  LoadMemory (CCA, WORD, pAddr, vAddr, DATA) GPR[rt] memword Exceptions: TLB Refill, TLB Invalid, Bus Error, Address Error, Watch LWC1 ILoad Word to Floating Point 31 26 25 21 20 16 15 0 655 16 Format: LWC1 ft, offset(base) MIPS32 Purpose: Load Word to Floating Point To load a word from memory to an FPR Description: FPR[ft]  memory[GPR[base] + offset] The contents of the 32-bit word at the memory location specified by the aligned effective address are fetched and placed into the low word of FPR ft. If FPRs are 64 bits wide, bits 63..32 of FPR ft become UNPREDICTABLE. The 16-bit signed offset is added to the contents of GPR base to form the effective address. Restrictions: Pre-Release 6: An Address Error exception occurs if EffectiveAddress1..0 ≠ 0 (not word-aligned). Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment. Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent. Operation: vAddr  sign_extend(offset) + GPR[base] (pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) memword  LoadMemory(CCA, WORD, pAddr, vAddr, DATA) StoreFPR(ft, UNINTERPRETED_WORD, memword) Exceptions: TLB Refill, TLB Invalid, Address Error, Reserved Instruction, Coprocessor Unusable, Watch LWC1 110001 base ft offset The MIPS32® Instruction Set Manual, Revision 6.06 236 LWC2 Load Word to Coprocessor 2 pre-Release 6 31 26 25 21 20 16 15 0 655 16 LWC2 110010 base rt offset Release 6 31 26 25 21 20 16 15 11 10 0 6 5 5 5 11 Format: LWC2 rt, offset(base) MIPS32 Purpose: Load Word to Coprocessor 2 To load a word from memory to a COP2 register. Description: CPR[2,rt,0]  memory[GPR[base] + offset] The contents of the 32-bit word at the memory location specified by the aligned effective address are fetched and placed into the low word of COP2 (Coprocessor 2) general register rt. The signed offset is added to the contents of GPR base to form the effective address. Restrictions: Pre-Release 6: An Address Error exception occurs if +EffectiveAddress1..0 ≠ 0 (not word-aligned). Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment. Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent. Availability and Compatibility This instruction has been recoded for Release 6. Operation: vAddr  sign_extend(offset) + GPR[base] (pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) memword  LoadMemory(CCA, DOUBLEWORD, pAddr, vAddr, DATA) CPR[2,rt,0]  memword Exceptions: TLB Refill, TLB Invalid, Address Error, Reserved Instruction, Coprocessor Unusable, Watch Programming Notes: Release 6 implements an 11-bit offset, whereas all release levels lower than Release 6 implement a 16-bit offset. COP2 010010 LWC2 01010 rt base offset 237 The MIPS32® Instruction Set Manual, Revision 6.06 LWC2 ILoad Word to Coprocessor 2 The MIPS32® Instruction Set Manual, Revision 6.06 238 LWE Load Word EVA 31 26 25 21 20 16 15 7 6 5 0 655916 Format: LWE rt, offset(base) MIPS32 Purpose: Load Word EVA To load a word from user mode virtual address space when executing in kernel mode. Description: GPR[rt]  memory[GPR[base] + offset] The contents of the 32-bit word at the memory location specified by the aligned effective address are fetched, sign- extended to the GPR register length if necessary, and placed in GPR rt. The 9-bit signed offset is added to the contents of GPR base to form the effective address. The LWE instruction functions the same as the LW instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment config- ured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Enhanced Virtual Addressing section for additional information. Implementation of this instruction is specified by the Config5EVA field being set to one. Restrictions: Only usable when access to Coprocessor0 is enabled and when accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode. Pre-Release 6: The effective address must be naturally-aligned. If either of the 2 least-significant bits of the address is non-zero, an Address Error exception occurs. Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment. Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent. Operation: vAddr  sign_extend(offset) + GPR[base] (pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) memword  LoadMemory (CCA, WORD, pAddr, vAddr, DATA) GPR[rt]  memword Exceptions: TLB Refill, TLB Invalid, Bus Error, Address Error, Watch, Reserved Instruction, Coprocessor Unusable SPECIAL3 011111 base rt offset 0 LWE 101111 239 The MIPS32® Instruction Set Manual, Revision 6.06 241 The MIPS32® Instruction Set Manual, Revision 6.06 LWL Load Word Left Restrictions: None Figure 4.2 Bytes Loaded by LWL Instruction Memory contents and byte offsets Initial contents of Dest Register most least — significance — 0123 big-endian offset (vAddr1. 0) I J K L e f g h 3 2 1 most 0 little-endian least — significance — Destination register contents after instruction (shaded is unchanged) Big-endian vAddr1..0 Little-endian 0 1 2 3 IJKL JKL h KL g h L f g h L fgh KL gh JKL h IJKL Availability and Compatibility: Release 6 removes the load/store-left/right family of instructions, and requires the system to support misaligned memory accesses. Operation: vAddr  sign_extend(offset) + GPR[base] (pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) pAddr  pAddr || (pAddr xor ReverseEndian2) PSIZE-1..2 if BigEndianMem = 0 then 1..0 pAddr  pAddr endif PSIZE-1..2 || 02 byte  vAddr xor BigEndianCPU2 1..0 memword  LoadMemory (CCA, byte, pAddr, vAddr, DATA) temp  memword7+8*byte..0 || GPR[rt]23-8*byte..0 GPR[rt]  temp Exceptions: TLB Refill, TLB Invalid, Bus Error, Address Error, Watch Programming Notes: The architecture provides no direct support for treating unaligned words as unsigned values, that is, zeroing bits 63..32 of the destination register when bit 31 is loaded. Historical Information: In the MIPS I architecture, the LWL and LWR instructions were exceptions to the load-delay scheduling restriction. A LWL or LWR instruction which was immediately followed by another LWL or LWR instruction, and used the same destination register would correctly merge the 1 to 4 loaded bytes with the data loaded by the previous instruc- tion. All such restrictions were removed from the architecture in MIPS II. 243 The MIPS32® Instruction Set Manual, Revision 6.06 LWLE Load Word Left EVA Restrictions: Figure 4.4 Bytes Loaded by LWLE Instruction Memory contents and byte offsets Initial contents of Dest Register most least — significance — 0123 big-endian offset (vAddr1. 0) I J K L e f g h 3 2 1 most 0 little-endian least — significance — Destination register contents after instruction (shaded is unchanged) Big-endian vAddr1..0 Little-endian 0 1 2 3 IJKL JKL h KL g h L f g h L fgh KL gh JKL h IJKL Only usable when access to Coprocessor0 is enabled and when accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode. Availability and Compatibility: Release 6 removes the load/store-left/right family of instructions, and requires the system to support misaligned memory accesses. Operation: vAddr  sign_extend(offset) + GPR[base] (pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) pAddr  pAddr || (pAddr xor ReverseEndian2) PSIZE-1..2 if BigEndianMem = 0 then 1..0 pAddr  pAddr endif PSIZE-1..2 || 02 byte  vAddr xor BigEndianCPU2 1..0 memword  LoadMemory (CCA, byte, pAddr, vAddr, DATA) temp  memword7+8*byte..0 || GPR[rt]23-8*byte..0 GPR[rt]  temp Exceptions: TLB Refill, TLB Invalid, Bus Error, Address Error, Watch, Reserved Instruction, Coprocessor Unusable Programming Notes: The architecture provides no direct support for treating unaligned words as unsigned values, that is, zeroing bits 63..32 of the destination register when bit 31 is loaded. Historical Information: In the MIPS I architecture, the LWL and LWR instructions were exceptions to the load-delay scheduling restriction. A LWL or LWR instruction which was immediately followed by another LWL or LWR instruction, and used the LWLE ILoad Word Left EVA same destination register would correctly merge the 1 to 4 loaded bytes with the data loaded by the previous instruc- tion. All such restrictions were removed from the architecture in MIPS II. The MIPS32® Instruction Set Manual, Revision 6.06 244 LWLE Load Word Left EVA 245 The MIPS32® Instruction Set Manual, Revision 6.06 LWPC 31 ILoad Word PC-relative 0 MIPS32 Release 6 26 25 21 20 19 18 652 19 Format: LWPC rs, offset Purpose: Load Word PC-relative To load a word from memory as a signed value, using a PC-relative address. Description: GPR[rs]  memory[ PC + sign_extend( offset << 2 ) ] PCREL 111011 rs LWPC 01 offset The offset is shifted left by 2 bits, sign-extended, and added to the address of the LWPC instruction. The contents of the 32-bit word at the memory location specified by the aligned effective address are fetched, sign- extended to the GPR register length if necessary, and placed in GPR rs. Restrictions: LWPC is naturally aligned, by specification. Availability and Compatibility: This instruction is introduced by and required as of Release 6. Operation vAddr  ( PC + sign_extend(offset)<<2) (pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) memword  LoadMemory (CCA, WORD, pAddr, vAddr, DATA) GPR[rs]  memword Exceptions: TLB Refill, TLB Invalid, TLB Read Inhibit, Bus Error, Address Error, Watch Programming Note The Release 6 PC-relative loads (LWPC) are considered data references. For the purposes of watchpoints (provided by the CP0 WatchHi and WatchLo registers) and EJTAG breakpoints, the PC-relative reference is considered to be a data reference rather than an instruction reference. That is, the watchpoint or breakpoint is triggered only if enabled for data references. The MIPS32® Instruction Set Manual, Revision 6.06 246 LWR ILoad Word Right Restrictions: None Figure 4.6 Bytes Loaded by LWR Instruction Memory contents and byte offsets Initial contents of Dest Register most least — significance— 0123 big-endian offset (vAddr1. 0) I J K L e f g h 3 2 1 most 0 little-endian least — significance — Destination register contents after instruction (shaded is unchanged) Big-endian vAddr1..0 Little-endian 0 1 2 3 efg I ef IJ e IJK IJKL IJKL e IJK e f IJ e f g I Availability and Compatibility: Release 6 removes the load/store-left/right family of instructions, and requires the system to support misaligned memory accesses. Operation: vAddr  sign_extend(offset) + GPR[base] (pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) pAddr  pAddr || (pAddr xor ReverseEndian2) PSIZE-1..2 if BigEndianMem = 0 then 1..0 pAddr  pAddr endif PSIZE-1..2 || 02 byte  vAddr xor BigEndianCPU2 1..0 memword  LoadMemory (CCA, byte, pAddr, vAddr, DATA) temp  memword31..32-8*byte || GPR[rt]31–8*byte..0 GPR[rt]  temp Exceptions: TLB Refill, TLB Invalid, Bus Error, Address Error, Watch Programming Notes: The architecture provides no direct support for treating unaligned words as unsigned values, that is, zeroing bits 63..32 of the destination register when bit 31 is loaded. Historical Information: In the MIPS I architecture, the LWL and LWR instructions were exceptions to the load-delay scheduling restriction. A LWL or LWR instruction which was immediately followed by another LWL or LWR instruction, and used the same destination register would correctly merge the 1 to 4 loaded bytes with the data loaded by the previous instruc- tion. All such restrictions were removed from the architecture in MIPS II. The MIPS32® Instruction Set Manual, Revision 6.06 248 LWR Load Word Right 249 The MIPS32® Instruction Set Manual, Revision 6.06 LWR ILoad Word Right The MIPS32® Instruction Set Manual, Revision 6.06 250 LWRE ILoad Word Right EVA The bytes loaded from memory to the destination register depend on both the offset of the effective address within an aligned word, that is, the low 2 bits of the address (vAddr1..0), and the current byte-ordering mode of the processor (big- or little-endian). The figure below shows the bytes loaded for every combination of offset and byte ordering. Figure 4.8 Bytes Loaded by LWRE Instruction Memory contents and byte offsets Initial contents of Dest Register most least — significance— 0123 big-endian offset (vAddr1. 0) I J K L e f g h 3 2 1 most 0 little-endian least — significance — Destination register contents after instruction (shaded is unchanged) Big-endian vAddr1..0 Little-endian 0 1 2 3 efg I ef IJ e IJK IJKL IJKL e IJK e f IJ e f g I Restrictions: Only usable when access to Coprocessor0 is enabled and when accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode. Availability and Compatibility: Release 6 removes the load/store-left/right family of instructions, and requires the system to support misaligned memory accesses. Operation: vAddr  sign_extend(offset) + GPR[base] (pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) pAddr  pAddr || (pAddr xor ReverseEndian2) PSIZE-1..2 if BigEndianMem = 0 then 1..0 pAddr  pAddr endif PSIZE-1..2 || 02 byte  vAddr xor BigEndianCPU2 1..0 memword  LoadMemory (CCA, byte, pAddr, vAddr, DATA) temp  memword31..32-8*byte || GPR[rt]31–8*byte..0 GPR[rt]  temp Exceptions: TLB Refill, TLB Invalid, Bus Error, Address Error, Watch, Reserved Instruction, Coprocessor Unusable Programming Notes: The architecture provides no direct support for treating unaligned words as unsigned values, that is, zeroing bits 63..32 of the destination register when bit 31 is loaded. The MIPS32® Instruction Set Manual, Revision 6.06 252 253 The MIPS32® Instruction Set Manual, Revision 6.06 LWRE Load Word Right EVA Historical Information: In the MIPS I architecture, the LWL and LWR instructions were exceptions to the load-delay scheduling restriction. A LWL or LWR instruction which was immediately followed by another LWL or LWR instruction, and used the same destination register would correctly merge the 1 to 4 loaded bytes with the data loaded by the previous instruc- tion. All such restrictions were removed from the architecture in MIPS II. LWXC1 ILoad Word Indexed to Floating Point 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: LWXC1 fd, index(base) MIPS32 Release 2, removed in Release 6 Purpose: Load Word Indexed to Floating Point To load a word from memory to an FPR (GPR+GPR addressing). Description: FPR[fd]  memory[GPR[base] + GPR[index]] The contents of the 32-bit word at the memory location specified by the aligned effective address are fetched and placed into the low word of FPR fd. If FPRs are 64 bits wide, bits 63..32 of FPR fs become UNPREDICTABLE. The contents of GPR index and GPR base are added to form the effective address. Restrictions: An Address Error exception occurs if EffectiveAddress1..0 ≠ 0 (not word-aligned). Availability and Compatibility: This instruction has been removed in Release 6. Required in all versions of MIPS64 since MIPS64 Release 1. Not available in MIPS32 Release 1. Required in MIPS32 Release 2 and all subsequent versions of MIPS32. When required, required whenever FPU is present, whether a 32-bit or 64-bit FPU, whether in 32-bit or 64-bit FP Register Mode (FIRF64=0 or 1, StatusFR=0 or 1). Operation: COP1X 010011 base index 0 00000 fd LWXC1 000000 vAddr  GPR[base] + GPR[index] if vAddr  02 then 1..0 SignalException(AddressError) endif (pAddr, CCA)  AddressTranslation (vAddr, DATA, LOAD) memword  LoadMemory(CCA, WORD, pAddr, vAddr, DATA) StoreFPR(fd, UNINTERPRETED_WORD, memword) Exceptions: TLB Refill, TLB Invalid, Address Error, Reserved Instruction, Coprocessor Unusable, Watch The MIPS32® Instruction Set Manual, Revision 6.06 254 MADD Multiply and Add Word to Hi, Lo 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: MADD rs, rt MIPS32, removed in Release 6 Purpose: Multiply and Add Word to Hi, Lo To multiply two words and add the result to Hi, Lo. Description: (HI,LO)  (HI,LO) + (GPR[rs] x GPR[rt]) The 32-bit word value in GPR rs is multiplied by the 32-bit word value in GPR rt, treating both operands as signed values, to produce a 64-bit result. The product is added to the 64-bit concatenated values of HI and LO. The most sig- nificant 32 bits of the result are written into HI and the least significant 32 bits are written into LO. No arithmetic exception occurs under any circumstances. Restrictions: This instruction does not provide the capability of writing directly to a target GPR. Availability and Compatibility: This instruction has been removed in Release 6. Operation: temp  (HI || LO) + (GPR[rs] x GPR[rt]) HI  temp63..32 LO  temp31..0 Exceptions: None Programming Notes: Where the size of the operands are known, software should place the shorter operand in GPR rt. This may reduce the latency of the instruction on those processors which implement data-dependent instruction latencies. SPECIAL2 011100 rs rt 0 0000 0 00000 MADD 000000 255 The MIPS32® Instruction Set Manual, Revision 6.06 MADD.fmt IFloating Point Multiply Add 31 26 25 21 20 16 15 11 10 6 5 3 2 0 6555533 COP1X 010011 fr ft fs fd MADD 100 fmt Format: MADD.fmt MADD.S fd, fr, fs, ft MADD.D fd, fr, fs, ft MADD.PS fd, fr, fs, ft Purpose: Floating Point Multiply Add To perform a combined multiply-then-add of FP values. Description: FPR[fd]  (FPR[fs] x FPR[ft])  FPR[fr] MIPS32 Release 2, removed in Release 6 MIPS32 Release 2, removed in Release 6 MIPS32 Release 2, removed in Release 6 The value in FPR fs is multiplied by the value in FPR ft to produce an intermediate product. The intermediate product is rounded according to the current rounding mode in FCSR. The value in FPR fr is added to the product. The result sum is calculated to infinite precision, rounded according to the current rounding mode in FCSR, and placed into FPR fd. The operands and result are values in format fmt. The results and flags are as if sepa- rate floating-point multiply and add instructions were executed. MADD.PS multiplies then adds the upper and lower halves of FPR fr, FPR fs, and FPR ft independently, and ORs together any generated exceptional conditions. The Cause bits are ORed into the Flag bits if no exception is taken. Restrictions: The fields fr, fs, ft, and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPREDICTABLE. The operands must be values in format fmt; if they are not, the result is UNPREDICTABLE and the value of the operand FPRs becomes UNPREDICTABLE. The result of MADD.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: MADD.S and MADD.D: Required in all versions of MIPS64 since MIPS64 Release 1. Not available in MIPS32 Release 1. Required in MIPS32 Release 2 and all subsequent versions of MIPS32. When required, these instructions are to be implemented if an FPU is present either in a 32-bit or 64-bit FPU or in a 32-bit or 64-bit FP Register Mode (FIRF64=0 or 1, StatusFR=0 or 1). This instruction has been removed in Release 6 and has been replaced by the fused multiply-add instruction. Refer to the fused multiply-add instruction ‘MADDF.fmt’ in this manual for more information. Release 6 does not support Paired Single (PS). Operation: vfr  ValueFPR(fr, fmt) vfs  ValueFPR(fs, fmt) vft  ValueFPR(ft, fmt) StoreFPR(fd, fmt, (vfs xfmt vft) fmt vfr) Exceptions: Coprocessor Unusable, Reserved Instruction The MIPS32® Instruction Set Manual, Revision 6.06 256 MADD.fmt Floating Point Multiply Add Floating Point Exceptions: Inexact, Unimplemented Operation, Invalid Operation, Overflow, Underflow 257 The MIPS32® Instruction Set Manual, Revision 6.06 MADD.fmt IFloating Point Multiply Add The MIPS32® Instruction Set Manual, Revision 6.06 258 MADDF.fmt MSUBF.fmt Floating Point Fused Multiply Add, Floating Point Fused Multiply Subtract 31 26 25 21 20 16 15 11 10 6 5 0 6555533 COP1 010001 fmt ft fs fd MADDF 011000 COP1 010001 fmt ft fs fd MSUBF 011001 259 The MIPS32® Instruction Set Manual, Revision 6.06 Format: MADDF.fmt MSUBF.fmt MADDF.S fd, fs, ft MADDF.D fd, fs, ft MSUBF.S fd, fs, ft MSUBF.D fd, fs, ft Purpose: Floating Point Fused Multiply Add, Floating Point Fused Multiply Subtract MADDF.fmt: To perform a fused multiply-add of FP values. MSUBF.fmt: To perform a fused multiply-subtract of FP values. Description: MADDF.fmt: FPR[fd]  FPR[fd] + (FPR[fs]  FPR[ft]) MSUBF.fmt: FPR[fd]  FPR[fd] - (FPR[fs]  FPR[ft]) MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 The value in FPR fs is multiplied by the value in FPR ft to produce an intermediate product. The intermediate product is calculated to infinite precision. The product is added to the value in FPR fd. The result sum is calculated to infinite precision, rounded according to the current rounding mode in FCSR, and placed into FPR fd. The operands and result are values in format fmt. (For MSUBF fmt, the product is subtracted from the value in FPR fd.) Cause bits are ORed into the Flag bits if no exception is taken. Restrictions: None Availability and Compatibility: MADDF.fmt and MSUBF.fmt are required in Release 6. MADDF.fmt and MSUBF.fmt are not available in architectures pre-Release 6. The fused multiply add instructions, MADDF.fmt and MSUBFfmt, replace pre-Release 6 instructions such as MADD fmt, MSUB.fmt, NMADD.fmt, and NMSUB.fmt. The replaced instructions were unfused multiply-add, with an intermediate rounding. Release 6 MSUBF fmt, fdfd-fsft, corresponds more closely to pre-Release 6 NMADD fmt, fdfr-fsft, than to pre-Release 6 MSUB.fmt, fdfsft-fr. FPU scalar MADDF fmt corresponds to MSA vector MADD.df. FPU scalar MSUBF fmt corresponds to MSA vector MSUB.df. Operation: if not IsCoprocessorEnabled(1) then SignalException(CoprocessorUnusable, 1) endif if not IsFloatingPointImplemented(fmt)) then SignalException(ReservedInstruction) endif MADDF.fmt MSUBF.fmt IFloating Point Fused Multiply Add, Floating Point Fused Multiply Subtract vfr  ValueFPR(fr, fmt) vfs  ValueFPR(fs, fmt) vfd  ValueFPR(fd, fmt) MADDF.fmt: vinf  vfd  (vfs * vft) MADDF.fmt: vinf  vfd - (vfs * vft) StoreFPR(fd, fmt, vinf) Special Considerations: The fused multiply-add computation is performed in infinite precision, and signals Inexact, Overflow, or Underflow if and only if the final result differs from the infinite precision result in the appropriate manner. Like most FPU computational instructions, if the flush-subnormals-to-zero mode, FCSR.FS=1, then subnormals are flushed before beginning the fused-multiply-add computation, and Inexact may be signaled. I.e. Inexact may be signaled both by input flushing and/or by the fused-multiply-add: the conditions or ORed. Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Inexact, Unimplemented Operation, Invalid Operation, Overflow, Underflow The MIPS32® Instruction Set Manual, Revision 6.06 260 MADDU Multiply and Add Unsigned Word to Hi,Lo 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: MADDU rs, rt MIPS32, removed in Release 6 Purpose: Multiply and Add Unsigned Word to Hi,Lo To multiply two unsigned words and add the result to HI, LO. Description: (HI,LO)  (HI,LO)  (GPR[rs] x GPR[rt]) The 32-bit word value in GPR rs is multiplied by the 32-bit word value in GPR rt, treating both operands as unsigned values, to produce a 64-bit result. The product is added to the 64-bit concatenated values of HI and LO. The most sig- nificant 32 bits of the result are written into HI and the least significant 32 bits are written into LO. No arithmetic exception occurs under any circumstances. Restrictions: None This instruction does not provide the capability of writing directly to a target GPR. Availability and Compatibility: This instruction has been removed in Release 6. Operation: temp  (HI || LO)  (GPR[rs] x GPR[rt]) HI  temp63..32 LO  temp31..0 Exceptions: None Programming Notes: Where the size of the operands are known, software should place the shorter operand in GPR rt. This may reduce the latency of the instruction on those processors which implement data-dependent instruction latencies. SPECIAL2 011100 rs rt 0 00000 0 00000 MADDU 000001 261 The MIPS32® Instruction Set Manual, Revision 6.06 MAX.fmt MIN.fmt MAXA.fmt MINA.fmt IScalar Floating-Point Max/Min/maxNumMag/minNumMag 31 31 31 31 26 25 21 20 16 15 11 10 6 5 0 655556 26 25 21 20 16 15 11 10 6 5 0 655556 26 25 21 20 16 15 11 10 6 5 0 655556 26 25 21 20 16 15 11 10 6 5 0 655556 COP1 010001 fmt ft fs fd MAX 011110 COP1 010001 fmt ft fs fd MAXA 011111 COP1 010001 fmt ft fs fd MIN 011100 COP1 010001 fmt ft fs fd MINA 011101 Format: MAX.fmt MIN.fmt MAX.S fd,fs,ft MAX.D fd,fs,ft MAXA.S fd,fs,ft MAXA.D fd,fs,ft MIN.S fd,fs,ft MIN.D fd,fs,ft MINA.S fd,fs,ft MINA.D fd,fs,ft MAXA.fmt MINA.fmt Purpose: Scalar Floating-Point Max/Min/maxNumMag/minNumMag Scalar Floating-Point Maximum Scalar Floating-Point Minimum Scalar Floating-Point argument with Maximum Absolute Value Scalar Floating-Point argument with Minimum Absolute Value Description: MAX.fmt: FPR[fd] maxNum(FPR[fs],FPR[ft]) MIN.fmt: FPR[fd] minNum(FPR[fs],FPR[ft]) MAXA.fmt: FPR[fd]maxNumMag(FPR[fs],FPR[ft]) MINA.fmt: FPR[fd]minNumMag(FPR[fs],FPR[ft]) MAX.fmt writes the maximum value of the inputs fs and ft to the destination fd. MIN.fmt writes the minimum value of the inputs fs and ft to the destination fd. MAXA fmt takes input arguments fs and ft and writes the argument with the maximum absolute value to the desti- nation fd. MINA fmt takes input arguments fs and ft and writes the argument with the minimum absolute value to the desti- nation fd. The instructions MAX.fmt/MIN fmt/MAXA fmt/MINA.fmt correspond to the IEEE 754-2008 operations maxNum/ The MIPS32® Instruction Set Manual, Revision 6.06 262 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MAX.fmt MIN.fmt MAXA.fmt MINA.fmt Scalar Floating-Point Max/Min/maxNumMag/minNumMag 263 The MIPS32® Instruction Set Manual, Revision 6.06 1. minNum/maxNumMag/minNumMag. • MAX.fmt corresponds to the IEEE 754-2008 operation maxNum. • MIN.fmt corresponds to the IEEE 754-2008 operation minNum. • MAXA fmt corresponds to the IEEE 754-2008 operation maxNumMag. • MINA fmt corresponds to the IEEE 754-2008 operation minNumMag. Numbers are preferred to NaNs: if one input is a NaN, but not both, the value of the numeric input is returned. If both are NaNs, the NaN in fs is returned.1 The scalar FPU instructions MAX fmt/MIN.fmt/MAXA fmt/MINA fmt correspond to the MSA instructions FMAX.df/FMIN.df/FMAXA.df/FMINA.df. • Scalar FPU instruction MAX fmt corresponds to the MSA vector instruction FMAX.df. • Scalar FPU instruction MIN fmt corresponds to the MSA vector instruction FMIN.df. • Scalar FPU instruction MAXA.fmt corresponds to the MSA vector instruction FMAX_A.df. • Scalar FPU instruction MINA.fmt corresponds to the MSA vector instruction FMIN_A.df. Restrictions: Data-dependent exceptions are possible as specified by the IEEE Standard for Floating-Point Arithmetic 754TM- 2008. See also the section “Special Cases”, below. Availability and Compatibility: These instructions are introduced by and required as of Release 6. Operation: if not IsCoprocessorEnabled(1) then SignalException(CoprocessorUnusable, 1) endif if not IsFloatingPointImplemented(fmt) then SignalException(ReservedInstruction) endif v1 ValueFPR(fs,fmt) v2 ValueFPR(ft,fmt) if SNaN(v1) or SNaN(v2) then then SignalException(InvalidOperand) jendifj if NaN(v1) and NaN(v2)then ftmp v1 elseif NaN(v1) then ftmp v2 elseif NaN(v2) then ftmp v1 else case instruction of IEEE standard 754-2008 allows either input to be chosen if both inputs are NaNs. Release 6 specifies that the first input must be propagated. MAX.fmt MIN.fmt MAXA.fmt MINA.fmt IScalar Floating-Point Max/Min/maxNumMag/minNumMag FMAX.fmt: ftmp  MaxFP.fmt(ValueFPR(fs,fmt),ValueFPR(ft,fmt)) FMIN.fmt: ftmp  MinFP.fmt(ValueFPR(fs,fmt),ValueFPR(ft,fmt)) FMAXA.fmt: ftmp MaxAbsoluteFP.fmt(ValueFPR(fs,fmt),ValueFPR(ft,fmt)) FMINA.fmt: ftmp MinAbsoluteFP.fmt(ValueFPR(fs,fmt),ValueFPR(ft,fmt)) end case endif StoreFPR (fd, fmt, ftmp) /* end of instruction */ function MaxFP(tt, ts, n) /* Returns the largest argument. */ endfunction MaxFP function MinFP(tt, ts, n) /* Returns the smallest argument. */ endfunction MaxFP function MaxAbsoluteFP(tt, ts, n) /* Returns the argument with largest absolute value. For equal absolute values, returns the largest argument.*/ endfunction MaxAbsoluteFP function MinAbsoluteFP(tt, ts, n) /* Returns the argument with smallest absolute value. For equal absolute values, returns the smallest argument.*/ endfunction MinAbsoluteFP function NaN(tt, ts, n) /* Returns true if the value is a NaN */ return SNaN(value) or QNaN(value) endfunction MinAbsoluteFP Table 4.1 Special Cases for FP MAX, MIN, MAXA, MINA Operand Other Release 6 Instructions fs ft MAX MIN MAXA MINA -0.0 0.0 0.0 -0.0 0.0 -0.0 0.0 -0.0 QNaN # # # # # # QNaN QNaN1 QNaN2 Release 6 QNan1 QNaN1 QNaN1 QNaN1 IEEE 754 2008 Arbitrary choice. Not allowed to clear sign bit. The MIPS32® Instruction Set Manual, Revision 6.06 264 MAX.fmt MIN.fmt MAXA.fmt MINA.fmt Scalar Floating-Point Max/Min/maxNumMag/minNumMag Table 4.1 Special Cases for FP MAX, MIN, MAXA, MINA Operand Other Release 6 Instructions fs ft MAX MIN MAXA MINA Either or both operands SNaN Invalid Operation exception enabled Signal Invalid Operation Exception. Destination not written. ... disabled Treat as if the SNaN were a QNaN (do not quieten the result). 265 The MIPS32® Instruction Set Manual, Revision 6.06 Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Unimplemented Operation, Invalid Operation MFC0 IMove from Coprocessor 0 31 26 25 21 20 16 15 11 10 3 2 655583 Format: MFC0 rt, rd MFC0 rt, rd, sel Purpose: Move from Coprocessor 0 To move the contents of a coprocessor 0 register to a general register. 0 MIPS32 MIPS32 COP0 010000 MF 00000 rt rd 0 00000000 sel Description: GPR[rt]  CPR[0,rd,sel] The contents of the coprocessor 0 register specified by the combination of rd and sel are loaded into general register rt. Not all coprocessor 0 registers support the sel field. In those instances, the sel field must be zero. Restrictions: Pre-Release 6: The results are UNDEFINED if coprocessor 0 does not contain a register as specified by rd and sel. Release 6: Reading a reserved register or a register that is not implemented for the current core configuration returns 0. Operation: reg = rd if IsCoprocessorRegisterImplemented(0, reg, sel) then data  CPR[0, reg, sel] GPR[rt]  data else if ArchitectureRevision() ≥ 6 then GPR[rt]  0 else UNDEFINED endif endif Exceptions: Coprocessor Unusable, Reserved Instruction The MIPS32® Instruction Set Manual, Revision 6.06 266 MFC1 Move Word From Floating Point 31 26 25 21 20 16 15 11 10 0 6 5 5 5 11 Format: MFC1 rt, fs MIPS32 Purpose: Move Word From Floating Point To copy a word from an FPU (CP1) general register to a GPR. Description: GPR[rt]  FPR[fs] The contents of FPR fs are loaded into general register rt. Restrictions: Operation: data  ValueFPR(fs, UNINTERPRETED_WORD) GPR[rt]  data Exceptions: Coprocessor Unusable, Reserved Instruction Historical Information: For MIPS I, MIPS II, and MIPS III the contents of GPR rt are UNPREDICTABLE for the instruction immediately following MFC1. COP1 010001 MF 00000 rt fs 0 000 0000 0000 267 The MIPS32® Instruction Set Manual, Revision 6.06 MFC2 IMove Word From Coprocessor 2 31 26 25 21 20 655 Format: MFC2 rt, Impl MFC2, rt, Impl, sel 16 15 11 10 8 7 0 MIPS32 MIPS32 COP2 010010 MF 00000 rt Impl The syntax shown above is an example using MFC1 as a model. The specific syntax is implementation dependent. Purpose: Move Word From Coprocessor 2 To copy a word from a COP2 general register to a GPR. Description: GPR[rt]  CP2CPR[Impl] The contents of the coprocessor 2 register denoted by the Impl field are and placed into general register rt. The inter- pretation of the Impl field is left entirely to the Coprocessor 2 implementation and is not specified by the architecture. Restrictions: The results are UNPREDICTABLE if the Impl field specifies a coprocessor 2 register that does not exist. Operation: data  CP2CPR[Impl] GPR[rt]  data Exceptions: Coprocessor Unusable The MIPS32® Instruction Set Manual, Revision 6.06 268 MFC2 Move Word From Coprocessor 2 269 The MIPS32® Instruction Set Manual, Revision 6.06 MFHC0 IMove from High Coprocessor 0 31 26 25 21 20 16 15 11 10 3 2 0 655583 Format: MFHC0 rt, rd MIPS32 Release 5 MFHC0 rt, rd, sel MIPS32 Release 5 Purpose: Move from High Coprocessor 0 To move the contents of the upper 32 bits of a Coprocessor 0 register, extended by 32-bits, to a general register. Description: GPR[rt]  CPR[0,rd,sel][63:32] The contents of the Coprocessor 0 register specified by the combination of rd and sel are loaded into general register rt. Not all Coprocessor 0 registers support the sel field, and in those instances, the sel field must be zero. The MFHC0 operation is not affected when the Coprocessor 0 register specified is the EntryLo0 or the EntryLo1 reg- ister. Data is read from the upper half of the 32-bit register extended to 64-bits without modification before writing to the GPR. This is because RI and XI bits are not repositioned on write from GPR to EntryLo0 or the EntryLo1. Restrictions: Pre-Release 6: The results are UNDEFINED if Coprocessor 0 does not contain a register as specified by rd and sel, or the register exists but is not extended by 32-bits,or the register is extended for XPA, but XPA is not supported or enabled. Release 6: Reading the high part of a register that is reserved, not implemented for the current core configuration, or that is not extended beyond 32 bits returns 0. Operation: if Config5MVH = 0 then SignalException(ReservedInstruction) endif reg  rd if IsCoprocessorRegisterImplemented(0, reg, sel) and IsCoprocessorRegisterExtended(0, reg, sel) then data  CPR[0, reg, sel] GPR[rt]  data63..32 else if ArchitectureRevision() ≥ 6 then GPR[rt]  0 else UNDEFINED endif endif Exceptions: Coprocessor Unusable, Reserved Instruction COP0 010000 MFH 00010 rt rd 0 00000000 sel The MIPS32® Instruction Set Manual, Revision 6.06 270 MFHC1 Move Word From High Half of Floating Point Register 31 26 25 21 20 16 15 11 10 6 5 5 5 11 Format: MFHC1 rt, fs Purpose: Move Word From High Half of Floating Point Register To copy a word from the high half of an FPU (CP1) general register to a GPR. 0 MIPS32 Release 2 COP1 010001 MFH 00011 rt fs 0 000 0000 0000 271 The MIPS32® Instruction Set Manual, Revision 6.06 Description: GPR[rt]  FPR[fs]63..32 The contents of the high word of FPR fs are loaded into general register rt. This instruction is primarily intended to support 64-bit floating point units on a 32-bit CPU, but the semantics of the instruction are defined for all cases. Restrictions: In implementations prior to Release 2 of the architecture, this instruction resulted in a Reserved Instruction exception. The results are UNPREDICTABLE if StatusFR = 0 and fs is odd. Operation: data  ValueFPR(fs, UNINTERPRETED_DOUBLEWORD)63..32 GPR[rt]  data Exceptions: Coprocessor Unusable, Reserved Instruction MFHC2 IMove Word From High Half of Coprocessor 2 Register 31 26 25 21 20 16 15 11 10 655 16 Format: MFHC2 rt, Impl MFHC2, rt, rd, sel 3 2 0 MIPS32 Release 2 MIPS32 Release 2 COP2 010010 MFH 00011 rt Impl The syntax shown above is an example using MFHC1 as a model. The specific syntax is implementation dependent. Purpose: Move Word From High Half of Coprocessor 2 Register To copy a word from the high half of a COP2 general register to a GPR. Description: GPR[rt]  CP2CPR[Impl]63..32 The contents of the high word of the coprocessor 2 register denoted by the Impl field are placed into GPR rt. The interpretation of the Impl field is left entirely to the Coprocessor 2 implementation and is not specified by the archi- tecture. Restrictions: The results are UNPREDICTABLE if the Impl field specifies a coprocessor 2 register that does not exist, or if that register is not 64 bits wide. In implementations prior to Release 2 of the architecture, this instruction resulted in a Reserved Instruction exception. Operation: data  CP2CPR[Impl]63..32 GPR[rt]  data Exceptions: Coprocessor Unusable, Reserved Instruction The MIPS32® Instruction Set Manual, Revision 6.06 272 MFHI Move From HI Register 31 26 25 16 15 11 10 6 5 0 6 10 5 5 6 SPECIAL 000000 0 00 0000 0000 rd 0 00000 MFHI 010000 273 The MIPS32® Instruction Set Manual, Revision 6.06 Format: MFHI rd Purpose: Move From HI Register To copy the special purpose HI register to a GPR. Description: GPR[rd]  HI The contents of special register HI are loaded into GPR rd. Restrictions: None Availability and Compatibility: This instruction has been removed in Release 6. Operation: GPR[rd]  HI Exceptions: None Historical Information: MIPS32, removed in Release 6 In the MIPS I, II, and III architectures, the two instructions which follow the MFHI must not modify the HI register. If this restriction is violated, the result of the MFHI is UNPREDICTABLE. This restriction was removed in MIPS IV and MIPS32, and all subsequent levels of the architecture. MFLO IMove From LO Register 31 26 25 16 15 11 10 6 5 0 6 10 5 5 6 Format: MFLO rd MIPS32, removed in Release 6 Purpose: Move From LO Register To copy the special purpose LO register to a GPR. Description: GPR[rd]  LO The contents of special register LO are loaded into GPR rd. Restrictions: None Availability and Compatibility: This instruction has been removed in Release 6. Operation: GPR[rd]  LO Exceptions: None Historical Information: In the MIPS I, II, and III architectures, the two instructions which follow the MFLO must not modify the HI register. If this restriction is violated, the result of the MFLO is UNPREDICTABLE. This restriction was removed in MIPS IV and MIPS32, and all subsequent levels of the architecture. SPECIAL 000000 0 00 0000 0000 rd 0 00000 MFLO 010010 The MIPS32® Instruction Set Manual, Revision 6.06 274 MOV.fmt Floating Point Move 31 26 25 21 20 16 15 11 10 6 5 655556 COP1 010001 fmt 0 00000 fs fd MOV 000110 275 The MIPS32® Instruction Set Manual, Revision 6.06 Format: MOV.fmt MOV.S fd, fs MOV.D fd, fs MOV.PS fd, fs Purpose: Floating Point Move To move an FP value between FPRs. 0 MIPS32 MIPS32 MIPS64,MIPS32 Release 2, removed in Release 6 Description: FPR[fd]  FPR[fs] The value in FPR fs is placed into FPR fd. The source and destination are values in format fmt. In paired-single for- mat, both the halves of the pair are copied to fd. The move is non-arithmetic; it causes no IEEE 754 exceptions, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPRE- DICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. The result of MOV.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: MOV.PS has been removed in Release 6. Operation: StoreFPR(fd, fmt, ValueFPR(fs, fmt)) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Unimplemented Operation MOVF IMove Conditional on Floating Point False 31 26 25 21 20 18 17 16 15 11 10 6 5 0 65311556 Format: MOVF rd, rs, cc MIPS32, removed in Release 6 Purpose: Move Conditional on Floating Point False To test an FP condition code then conditionally move a GPR. Description: if FPConditionCode(cc) = 0 then GPR[rd]  GPR[rs] If the floating point condition code specified by CC is zero, then the contents of GPR rs are placed into GPR rd. Restrictions: Availability and Compatibility: This instruction has been removed in Release 6. Operation: if FPConditionCode(cc) = 0 then GPR[rd]  GPR[rs] endif Exceptions: Reserved Instruction, Coprocessor Unusable SPECIAL 000000 rs cc 0 0 tf 0 rd 0 00000 MOVCI 000001 The MIPS32® Instruction Set Manual, Revision 6.06 276 MOVF.fmt Floating Point Move Conditional on Floating Point False 31 26 25 21 20 18 17 16 15 11 10 6 5 0 65311556 COP1 010001 fmt cc 0 0 tf 0 fs fd MOVCF 010001 277 The MIPS32® Instruction Set Manual, Revision 6.06 Format: MOVF.fmt MOVF.S fd, fs, cc MOVF.D fd, fs, cc MOVF.PS fd, fs, cc Purpose: Floating Point Move Conditional on Floating Point False To test an FP condition code then conditionally move an FP value. MIPS32, removed in Release 6 MIPS32, removed in Release 6 removed in Release 6 Description: if FPConditionCode(cc) = 0 then FPR[fd]  FPR[fs] If the floating point condition code specified by CC is zero, then the value in FPR fs is placed into FPR fd. The source and destination are values in format fmt. If the condition code is not zero, then FPR fs is not copied and FPR fd retains its previous value in format fmt. If fd did not contain a value either in format fmt or previously unused data from a load or move-to operation that could be interpreted in format fmt, then the value of fd becomes UNPREDICTABLE. MOVF.PS merges the lower half of FPR fs into the lower half of FPR fd if condition code CC is zero, and indepen- dently merges the upper half of FPR fs into the upper half of FPR fd if condition code CC+1 is zero. The CC field must be even; if it is odd, the result of this operation is UNPREDICTABLE. The move is non-arithmetic; it causes no IEEE 754 exceptions, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPRE- DICTABLE. The operand must be a value in format fmt. If it is not, the result is UNPREDITABLE and the value of the operand FPR becomes UNPREDICTABLE. The result of MOVF.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model; it is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: This instruction has been removed in Release 6 and has been replaced by the ‘SEL.fmt’ instruction. Refer to the SEL fmt instruction in this manual for more information. Release 6 does not support Paired Single (PS). Operation: if FPConditionCode(cc) = 0 then StoreFPR(fd, fmt, ValueFPR(fs, fmt)) else StoreFPR(fd, fmt, ValueFPR(fd, fmt)) Exceptions: Coprocessor Unusable, Reserved Instruction MOVF.fmt IFloating Point Move Conditional on Floating Point False Floating Point Exceptions: Unimplemented Operation The MIPS32® Instruction Set Manual, Revision 6.06 278 MOVN Move Conditional on Not Zero 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: MOVN rd, rs, rt MIPS32, removed in Release 6 Purpose: Move Conditional on Not Zero To conditionally move a GPR after testing a GPR value. Description: if GPR[rt]  0 then GPR[rd]  GPR[rs] If the value in GPR rt is not equal to zero, then the contents of GPR rs are placed into GPR rd. Restrictions: None Availability and Compatibility: This instruction has been removed in Release 6 and has been replaced by the ‘SELNEZ’ instruction. Refer to the SELNEZ instruction in this manual for more information. Operation: if GPR[rt]  0 then GPR[rd]  GPR[rs] endif Exceptions: None Programming Notes: The non-zero value tested might be the condition true result from the SLT, SLTI, SLTU, and SLTIU comparison instructions or a boolean value read from memory. SPECIAL 000000 rs rt rd 0 00000 MOVN 001011 279 The MIPS32® Instruction Set Manual, Revision 6.06 MOVN.fmt IFloating Point Move Conditional on Not Zero 31 26 25 21 20 16 15 11 10 6 5 0 655556 COP1 010001 fmt rt fs fd MOVN 010011 Format: MOVN.fmt MOVN.S fd, fs, rt MOVN.D fd, fs, rt MOVN.PS fd, fs, rt Purpose: Floating Point Move Conditional on Not Zero To test a GPR then conditionally move an FP value. MIPS32, removed in Release 6 MIPS32, removed in Release 6 MIPS32 Release 2, removed in Release 6 Description: if GPR[rt]  0 then FPR[fd]  FPR[fs] If the value in GPR rt is not equal to zero, then the value in FPR fs is placed in FPR fd. The source and destination are values in format fmt. If GPR rt contains zero, then FPR fs is not copied and FPR fd contains its previous value in format fmt. If fd did not contain a value either in format fmt or previously unused data from a load or move-to operation that could be inter- preted in format fmt, then the value of fd becomes UNPREDICTABLE. The move is non-arithmetic; it causes no IEEE 754 exceptions, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPRE- DICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. The result of MOVN.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: This instruction has been removed in Release 6 and has been replaced by the ‘SELNEZ fmt’ instruction. Refer to the SELNEZ.fmt instruction in this manual for more information. Release 6 does not support Paired Single (PS). Operation: if GPR[rt]  0 then StoreFPR(fd, fmt, ValueFPR(fs, fmt)) else StoreFPR(fd, fmt, ValueFPR(fd, fmt)) endif Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Unimplemented Operation The MIPS32® Instruction Set Manual, Revision 6.06 280 MOVT Move Conditional on Floating Point True 31 26 25 21 20 18 17 16 15 11 10 6 5 0 65311556 Format: MOVT rd, rs, cc MIPS32, removed in Release 6 Purpose: Move Conditional on Floating Point True To test an FP condition code then conditionally move a GPR. Description: if FPConditionCode(cc) = 1 then GPR[rd]  GPR[rs] If the floating point condition code specified by CC is one, then the contents of GPR rs are placed into GPR rd. Restrictions: Availability and Compatibility: This instruction has been removed in Release 6. Operation: if FPConditionCode(cc) = 1 then GPR[rd]  GPR[rs] endif Exceptions: Reserved Instruction, Coprocessor Unusable SPECIAL 000000 rs cc 0 0 tf 1 rd 0 00000 MOVCI 000001 281 The MIPS32® Instruction Set Manual, Revision 6.06 MOVT.fmt IFloating Point Move Conditional on Floating Point True 31 26 25 21 20 18 17 16 15 11 10 6 5 0 65311556 COP1 010001 fmt cc 0 0 tf 1 fs fd MOVCF 010001 Format: MOVT.fmt MOVT.S fd, fs, cc MOVT.D fd, fs, cc MOVT.PS fd, fs, cc Purpose: Floating Point Move Conditional on Floating Point True To test an FP condition code then conditionally move an FP value. MIPS32, removed in Release 6 MIPS32, removed in Release 6 MIPS32 Release 2, removed in Release 6 Description: if FPConditionCode(cc) = 1 then FPR[fd]  FPR[fs] If the floating point condition code specified by CC is one, then the value in FPR fs is placed into FPR fd. The source and destination are values in format fmt. If the condition code is not one, then FPR fs is not copied and FPR fd contains its previous value in format fmt. If fd did not contain a value either in format fmt or previously unused data from a load or move-to operation that could be interpreted in format fmt, then the value of fd becomes UNPREDICTABLE. MOVT.PS merges the lower half of FPR fs into the lower half of FPR fd if condition code CC is one, and indepen- dently merges the upper half of FPR fs into the upper half of FPR fd if condition code CC+1 is one. The CC field should be even; if it is odd, the result of this operation is UNPREDICTABLE. The move is non-arithmetic; it causes no IEEE 754 exceptions, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPRE- DICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. The result of MOVT.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility This instruction has been removed in Release 6 and has been replaced by the ‘SEL.fmt’ instruction. Refer to the SEL fmt instruction in this manual for more information. Release 6 does not support Paired Single (PS). Operation: if FPConditionCode(cc) = 1 then StoreFPR(fd, fmt, ValueFPR(fs, fmt)) else StoreFPR(fd, fmt, ValueFPR(fd, fmt)) endif Exceptions: Coprocessor Unusable, Reserved Instruction The MIPS32® Instruction Set Manual, Revision 6.06 282 283 The MIPS32® Instruction Set Manual, Revision 6.06 MOVT.fmt Floating Point Move Conditional on Floating Point True Floating Point Exceptions: Unimplemented Operation MOVZ IMove Conditional on Zero 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: MOVZ rd, rs, rt MIPS32, removed in Release 6 Purpose: Move Conditional on Zero To conditionally move a GPR after testing a GPR value. Description: if GPR[rt] = 0 then GPR[rd]  GPR[rs] If the value in GPR rt is equal to zero, then the contents of GPR rs are placed into GPR rd. Restrictions: None Availability and Compatibility: This instruction has been removed in Release 6 and has been replaced by the ‘SELEQZ’ instruction. Refer to the SELEQZ instruction in this manual for more information. Operation: if GPR[rt] = 0 then GPR[rd]  GPR[rs] endif Exceptions: None Programming Notes: The zero value tested might be the condition false result from the SLT, SLTI, SLTU, and SLTIU comparison instruc- tions or a boolean value read from memory. SPECIAL 000000 rs rt rd 0 00000 MOVZ 001010 The MIPS32® Instruction Set Manual, Revision 6.06 284 MOVZ.fmt Floating Point Move Conditional on Zero 31 26 25 21 20 16 15 11 10 6 5 0 655556 COP1 010001 fmt rt fs fd MOVZ 010010 285 The MIPS32® Instruction Set Manual, Revision 6.06 Format: MOVZ.fmt MOVZ.S fd, fs, rt MOVZ.D fd, fs, rt MOVZ.PS fd, fs, rt Purpose: Floating Point Move Conditional on Zero To test a GPR then conditionally move an FP value. MIPS32, removed in Release 6 MIPS32, removed in Release 6 MIPS32 Release 2, removed in Release 6 Description: if GPR[rt] = 0 then FPR[fd]  FPR[fs] If the value in GPR rt is equal to zero then the value in FPR fs is placed in FPR fd. The source and destination are val- ues in format fmt. If GPR rt is not zero, then FPR fs is not copied and FPR fd contains its previous value in format fmt. If fd did not con- tain a value either in format fmt or previously unused data from a load or move-to operation that could be interpreted in format fmt, then the value of fd becomes UNPREDICTABLE. The move is non-arithmetic; it causes no IEEE 754 exceptions, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPRE- DICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. The result of MOVZ.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: This instruction has been removed in Release 6 and has been replaced by the ‘SELEQZ fmt’ instruction. Refer to the SELEQZ.fmt instruction in this manual for more information. Release 6 does not support Paired Single (PS). Operation: if GPR[rt] = 0 then StoreFPR(fd, fmt, ValueFPR(fs, fmt)) else StoreFPR(fd, fmt, ValueFPR(fd, fmt)) endif Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Unimplemented Operation MSUB IMultiply and Subtract Word to Hi, Lo 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: MSUB rs, rt MIPS32, removed in Release 6 Purpose: Multiply and Subtract Word to Hi, Lo To multiply two words and subtract the result from HI, LO. Description: (HI,LO)  (HI,LO) - (GPR[rs] x GPR[rt]) The 32-bit word value in GPR rs is multiplied by the 32-bit value in GPR rt, treating both operands as signed values, to produce a 64-bit result. The product is subtracted from the 64-bit concatenated values of HI and LO. The most sig- nificant 32 bits of the result are written into HI and the least significant 32 bits are written into LO. No arithmetic exception occurs under any circumstances. Restrictions: No restrictions in any architecture releases except Release 6. This instruction does not provide the capability of writing directly to a target GPR. Availability and Compatibility: This instruction has been removed in Release 6. Operation: temp  (HI || LO) - (GPR[rs] x GPR[rt]) HI  temp63..32 LO  temp31..0 Exceptions: None Programming Notes: Where the size of the operands are known, software should place the shorter operand in GPR rt. This may reduce the latency of the instruction on those processors which implement data-dependent instruction latencies. SPECIAL2 011100 rs rt 0 00000 0 00000 MSUB 000100 The MIPS32® Instruction Set Manual, Revision 6.06 286 MSUB.fmt Floating Point Multiply Subtract 31 26 25 21 20 16 15 11 10 6 5 3 2 0 6555533 COP1X 010011 fr ft fs fd MSUB 101 fmt 287 The MIPS32® Instruction Set Manual, Revision 6.06 Format: MSUB.fmt MSUB.S fd, fr, fs, ft MSUB.D fd, fr, fs, ft MSUB.PS fd, fr, fs, ft Purpose: Floating Point Multiply Subtract To perform a combined multiply-then-subtract of FP values. Description: FPR[fd]  (FPR[fs] x FPR[ft])  FPR[fr] MIPS32 Release 2, removed in Release 6 MIPS32 Release 2, removed in Release 6 MIPS32 Release 2, removed in Release 6 The value in FPR fs is multiplied by the value in FPR ft to produce an intermediate product. The intermediate product is rounded according to the current rounding mode in FCSR. The subtraction result is calculated to infinite precision, rounded according to the current rounding mode in FCSR, and placed into FPR fd. The operands and result are values in format fmt. The results and flags are as if separate floating-point multiply and subtract instructions were executed. MSUB.PS multiplies then subtracts the upper and lower halves of FPR fr, FPR fs, and FPR ft independently, and ORs together any generated exceptional conditions. The Cause bits are ORed into the Flag bits if no exception is taken. Restrictions: The fields fr, fs, ft, and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPREDICTABLE. The operands must be values in format fmt; if they are not, the result is UNPREDICTABLE and the value of the operand FPRs becomes UNPREDICTABLE. The result of MSUB.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: MSUB.S and MSUB.D: Required in all versions of MIPS64 since MIPS64 Release 1. Not available in MIPS32 Release 1. Required in MIPS32 Release 2 and all subsequent versions of MIPS32. When required, these instructions are to be implemented if an FPU is present, either in a 32-bit or 64-bit FPU or in a 32-bit or 64-bit FP Register Mode (FIRF64=0 or 1, StatusFR=0 or 1). This instruction has been removed in Release 6 and has been replaced by the fused multiply-subtract instruction. Refer to the fused multiply-subtract instruction ‘MSUBF.fmt’ in this manual for more information. Release 6 does not support Paired Single (PS). Operation: vfr  ValueFPR(fr, fmt) vfs  ValueFPR(fs, fmt) vft  ValueFPR(ft, fmt) StoreFPR(fd, fmt, (vfs xfmt vft) fmt vfr)) Exceptions: Coprocessor Unusable, Reserved Instruction MSUB.fmt IFloating Point Multiply Subtract Floating Point Exceptions: Inexact, Unimplemented Operation, Invalid Operation, Overflow, Underflow The MIPS32® Instruction Set Manual, Revision 6.06 288 MSUBU Multiply and Subtract Word to Hi,Lo 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: MSUBU rs, rt MIPS32, removed in Release 6 Purpose: Multiply and Subtract Word to Hi,Lo To multiply two words and subtract the result from HI, LO. Description: (HI,LO)  (HI,LO)  (GPR[rs] x GPR[rt]) The 32-bit word value in GPR rs is multiplied by the 32-bit word value in GPR rt, treating both operands as unsigned values, to produce a 64-bit result. The product is subtracted from the 64-bit concatenated values of HI and LO. The most significant 32 bits of the result are written into HI and the least significant 32 bits are written into LO. No arith- metic exception occurs under any circumstances. Restrictions: This instruction does not provide the capability of writing directly to a target GPR. Availability and Compatibility: This instruction has been removed in Release 6. Operation: temp  (HI || LO) - (GPR[rs]  GPR[rt]) HI  temp63..32 LO  temp31..0 Exceptions: None Programming Notes: Where the size of the operands are known, software should place the shorter operand in GPR rt. This may reduce the latency of the instruction on those processors which implement data-dependent instruction latencies. SPECIAL2 011100 rs rt 0 00000 0 00000 MSUBU 000101 289 The MIPS32® Instruction Set Manual, Revision 6.06 MTC0 IMove to Coprocessor 0 31 26 25 21 20 16 15 11 10 3 2 655583 Format: MTC0 rt, rd MTC0 rt, rd, sel Purpose: Move to Coprocessor 0 To move the contents of a general register to a coprocessor 0 register. 0 MIPS32 MIPS32 COP0 010000 MT 00100 rt rd 0 0000 000 sel Description: CPR[0, rd, sel]  GPR[rt] The contents of general register rt are loaded into the coprocessor 0 register specified by the combination of rd and sel. Not all coprocessor 0 registers support the sel field. In those instances, the sel field must be set to zero. In Release 5, for a 32-bit processor, the MTC0 instruction writes all zeroes to the high-order bits of selected CP0 reg- isters that have been extended beyond 32 bits. This is required for compatibility with legacy software that does not use MTHC0, yet has hardware support for extended CP0 registers (such as for Extended Physical Addressing (XPA)). Because MTC0 overwrites the result of MTHC0, software must first read the high-order bits before writing the low- order bits, then write the high-order bits back either modified or unmodified. For initialization of an extended register, software may first write the low-order bits, then the high-order bits, without first reading the high-order bits. Restrictions: Pre-Release 6: The results are UNDEFINED if coprocessor 0 does not contain a register as specified by rd and sel. Release 6: Writes to a register that is reserved or not defined for the current core configuration are ignored. Operation: data  GPR[rt] regrd if IsCoprocessorRegisterImplemented (0, reg, sel) then CPR[0,reg,sel]  data if (Config5XPA = 1) then // The most-significant bit may vary by register. Only supported // bits should be written 0. Extended LLAddr is not written with 0s, // as it is a read-only register. BadVAddr is not written with 0s, as // it is read-only if (Config3LPA = 1) then if (reg,sel = EntryLo0 or EntryLo1) then CPR[0,reg,sel]63:32 = 032 endif if (reg,sel = MAAR) then CPR[0,reg,sel]63:32 = 032 endif // TagLo is zeroed only if the implementation-dependent bits // are writeable if (reg,sel = TagLo) then CPR[0,reg,sel]63:32 = 032 endif if (Config3VZ = 1) then if (reg,sel = EntryHi) then CPR[0,reg,sel]63:32 = 032 endif endif endif endif else if ArchitectureRevision() ≥ 6 then // nop (no exceptions, coprocessor state not modified) else UNDEFINED The MIPS32® Instruction Set Manual, Revision 6.06 290 291 The MIPS32® Instruction Set Manual, Revision 6.06 MTC0 Move to Coprocessor 0 endif endif Exceptions: Coprocessor Unusable, Reserved Instruction MTC1 IMove Word to Floating Point 31 26 25 21 20 16 15 11 10 0 6 5 5 5 11 Format: MTC1 rt, fs MIPS32 Purpose: Move Word to Floating Point To copy a word from a GPR to an FPU (CP1) general register. Description: FPR[fs]  GPR[rt] The low word in GPR rt is placed into the low word of FPR fs. Restrictions: Operation: data  GPR[rt]31..0 StoreFPR(fs, UNINTERPRETED_WORD, data) Exceptions: Coprocessor Unusable Historical Information: For MIPS I, MIPS II, and MIPS III the value of FPR fs is UNPREDICTABLE for the instruction immediately fol- lowing MTC1. COP1 010001 MT 00100 rt fs 0 000 0000 0000 The MIPS32® Instruction Set Manual, Revision 6.06 292 MTC2 Move Word to Coprocessor 2 31 26 25 21 20 16 15 11 10 8 7 655 16 Format: MTC2 rt, Impl MTC2 rt, Impl, sel 0 MIPS32 MIPS32 COP2 010010 MT 00100 rt Impl 293 The MIPS32® Instruction Set Manual, Revision 6.06 The syntax shown above is an example using MTC1 as a model. The specific syntax is implementation-dependent. Purpose: Move Word to Coprocessor 2 To copy a word from a GPR to a COP2 general register. Description: CP2CPR[Impl]  GPR[rt] The low word in GPR rt is placed into the low word of a Coprocessor 2 general register denoted by the Impl field. The interpretation of the Impl field is left entirely to the Coprocessor 2 implementation and is not specified by the architecture. Restrictions: The results are UNPREDICTABLE if the Impl field specifies a Coprocessor 2 register that does not exist. Operation: data  GPR[rt] CP2CPR[Impl]  data Exceptions: Coprocessor Unusable, Reserved Instruction MTHC0 IMove to High Coprocessor 0 31 26 25 21 20 16 15 11 10 3 2 0 655583 Format: MTHC0 rt, rd MIPS32 Release 5 MTHC0 rt, rd, sel MIPS32 Release 5 Purpose: Move to High Coprocessor 0 To copy a word from a GPR to the upper 32 bits of a CP0 general register that has been extended by 32 bits. Description: CPR[0, rd, sel][63:32]  GPR[rt] The contents of general register rt are loaded into the Coprocessor 0 register specified by the combination of rd and sel. Not all Coprocessor 0 registers support the sel field; the sel field must be set to zero. Restrictions: Pre-Release 6: The results are UNDEFINED if Coprocessor 0 does not contain a register as specified by rd and sel, or if the register exists but is not extended by 32 bits, or the register is extended for XPA, but XPA is not supported or enabled. Release 6: A write to the high part of a register that is reserved, not implemented for the current core, or that is not extended beyond 32 bits is ignored. Operation: if Config5MVH = 0 then SignalException(ReservedInstruction) endif data  GPR[rt] regrd if IsCoprocessorRegisterImplemented (0, reg, sel) and IsCoprocessorRegisterExtended (0, reg, sel) then CPR[0, reg, sel][63:32]  data else if ArchitectureRevision() ≥ 6 then // nop (no exceptions, coprocessor state not modified) else UNDEFINED endif endif Exceptions: Coprocessor Unusable, Reserved Instruction COP0 010000 MTH 00110 rt rd 0 0000 0000 sel The MIPS32® Instruction Set Manual, Revision 6.06 294 MTHC1 Move Word to High Half of Floating Point Register 31 26 25 21 20 16 15 11 10 6 5 5 5 11 Format: MTHC1 rt, fs Purpose: Move Word to High Half of Floating Point Register To copy a word from a GPR to the high half of an FPU (CP1) general register. 0 MIPS32 Release 2 COP1 010001 MTH 00111 rt fs 0 000 0000 0000 295 The MIPS32® Instruction Set Manual, Revision 6.06 Description: FPR[fs]63..32  GPR[rt] The word in GPR rt is placed into the high word of FPR fs. This instruction is primarily intended to support 64-bit floating point units on a 32-bit CPU, but the semantics of the instruction are defined for all cases. Restrictions: In implementations prior to Release 2 of the architecture, this instruction resulted in a Reserved Instruction exception. The results are UNPREDICTABLE if StatusFR = 0 and fs is odd. Operation: newdata  GPR[rt] olddata  ValueFPR(fs, UNINTERPRETED_DOUBLEWORD)31..0 StoreFPR(fs, UNINTERPRETED_DOUBLEWORD, newdata || olddata) Exceptions: Coprocessor Unusable, Reserved Instruction Programming Notes When paired with MTC1 to write a value to a 64-bit FPR, the MTC1 must be executed first, followed by the MTHC1. This is because of the semantic definition of MTC1, which is not aware that software is using an MTHC1 instruction to complete the operation, and sets the upper half of the 64-bit FPR to an UNPREDICTABLE value. MTHC2 IMove Word to High Half of Coprocessor 2 Register 31 26 25 21 20 16 15 11 10 655 16 Format: MTHC2 rt, Impl MTHC2 rt, Impl, sel 0 MIPS32 Release 2 MIPS32 Release 2 COP2 010010 MTH 00111 rt Impl The syntax shown above is an example using MTHC1 as a model. The specific syntax is implementation dependent. Purpose: Move Word to High Half of Coprocessor 2 Register To copy a word from a GPR to the high half of a COP2 general register. Description: CP2CPR[Impl]63..32  GPR[rt] The word in GPR rt is placed into the high word of coprocessor 2 general register denoted by the Impl field. The interpretation of the Impl field is left entirely to the Coprocessor 2 implementation and is not specified by the archi- tecture. Restrictions: The results are UNPREDICTABLE if the Impl field specifies a coprocessor 2 register that does not exist, or if that register is not 64 bits wide. In implementations prior to Release 2 of the architecture, this instruction resulted in a Reserved Instruction exception. Operation: data  GPR[rt] CP2CPR[Impl]  data || CPR[2,rd,sel]31..0 Exceptions: Coprocessor Unusable, Reserved Instruction Programming Notes When paired with MTC2 to write a value to a 64-bit CPR, the MTC2 must be executed first, followed by the MTHC2. This is because of the semantic definition of MTC2, which is not aware that software is using an MTHC2 instruction to complete the operation, and sets the upper half of the 64-bit CPR to an UNPREDICTABLE value. The MIPS32® Instruction Set Manual, Revision 6.06 296 MTHI Move to HI Register 31 26 25 21 20 6 5 0 6 5 15 6 Format: MTHI rs MIPS32, removed in Release 6 Purpose: Move to HI Register To copy a GPR to the special purpose HI register. Description: HI  GPR[rs] The contents of GPR rs are loaded into special register HI. Restrictions: A computed result written to the HI/LO pair by DIV, DIVU, MULT, or MULTU must be read by MFHI or MFLO before a new result can be written into either HI or LO. If an MTHI instruction is executed following one of these arithmetic instructions, but before an MFLO or MFHI instruction, the contents of LO are UNPREDICTABLE. The following example shows this illegal situation: MULT r2,r4 # start operation that will eventually write to HI,LO # code not containing mfhi or mflo # code not containing mflo # this mflo would get an UNPREDICTABLE value Availability and Compatibility: This instruction has been removed in Release 6. Operation: HI  GPR[rs] Exceptions: None Historical Information: In MIPS I-III, if either of the two preceding instructions is MFHI, the result of that MFHI is UNPREDICTABLE. Reads of the HI or LO special register must be separated from any subsequent instructions that write to them by two or more instructions. In MIPS IV and later, including MIPS32, this restriction does not exist. SPECIAL 000000 rs 0 000 0000 0000 0000 MTHI 010001 297 The MIPS32® Instruction Set Manual, Revision 6.06 ... MTHI r6 ... MFLO r3 MTLO IMove to LO Register 31 26 25 21 20 6 5 0 6 5 15 6 Format: MTLO rs MIPS32, removed in Release 6 Purpose: Move to LO Register To copy a GPR to the special purpose LO register. Description: LO  GPR[rs] The contents of GPR rs are loaded into special register LO. Restrictions: A computed result written to the HI/LO pair by DIV, DIVU, MULT, or MULTU must be read by MFHI or MFLO before a new result can be written into either HI or LO. If an MTLO instruction is executed following one of these arithmetic instructions, but before an MFLO or MFHI instruction, the contents of HI are UNPREDICTABLE. The following example shows this illegal situation: SPECIAL 000000 rs 0 000 0000 0000 0000 MTLO 010011 MULT ... MTLO ... MFHI r2,r4 # start operation that will eventually write to HI,LO # code not containing mfhi or mflo r6 r3 # code not containing mfhi # this mfhi would get an UNPREDICTABLE value Availability and Compatibility: This instruction has been removed in Release 6. Operation: LO  GPR[rs] Exceptions: None Historical Information: In MIPS I-III, if either of the two preceding instructions is MFHI, the result of that MFHI is UNPREDICTABLE. Reads of the HI or LO special register must be separated from any subsequent instructions that write to them by two or more instructions. In MIPS IV and later, including MIPS32, this restriction does not exist. The MIPS32® Instruction Set Manual, Revision 6.06 298 MUL IMultiply Word to GPR 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: MUL rd, rs, rt MIPS32, removed in Release 6 Purpose: Multiply Word to GPR To multiply two words and write the result to a GPR. Description: GPR[rd]  GPR[rs] x GPR[rt] The 32-bit word value in GPR rs is multiplied by the 32-bit value in GPR rt, treating both operands as signed values, to produce a 64-bit result. The least significant 32 bits of the product are written to GPR rd. The contents of HI and LO are UNPREDICTABLE after the operation. No arithmetic exception occurs under any circumstances. Restrictions: Note that this instruction does not provide the capability of writing the result to the HI and LO registers. Availability and Compatibility: The pre-Release 6 MUL instruction has been removed in Release 6. It has been replaced by a similar instruction of the same mnemonic, MUL, but different encoding, which is a member of a family of single-width multiply instruc- tions. Refer to the ‘MUL’ and ‘MUH’ instructions in this manual for more information. Operation: temp  GPR[rs] x GPR[rt] GPR[rd]  temp31..0 HI  UNPREDICTABLE LO  UNPREDICTABLE Exceptions: None Programming Notes: In some processors the integer multiply operation may proceed asynchronously and allow other CPU instructions to execute before it is complete. An attempt to read GPR rd before the results are written interlocks until the results are ready. Asynchronous execution does not affect the program result, but offers an opportunity for performance improvement by scheduling the multiply so that other instructions can execute in parallel. Programs that require overflow detection must check for it explicitly. Where the size of the operands are known, software should place the shorter operand in GPR rt. This may reduce the latency of the instruction on those processors which implement data-dependent instruction latencies. SPECIAL2 011100 rs rt rd 0 00000 MUL 000010 The MIPS32® Instruction Set Manual, Revision 6.06 299 MUL MUH MULU MUHU IMultiply Integers (with result to GPR) 31 26 25 21 20 16 15 11 10 6 5 0 SPECIAL 000000 rs rt rd MUL 00010 SOP30 011000 SPECIAL 000000 rs rt rd MUH 00011 SOP30 011000 SPECIAL 000000 rs rt rd MULU 00010 SOP31 011001 SPECIAL 000000 rs rt rd MUHU 00011 SOP31 011001 655556 Format: MUL MUH MULU MUHU MUL rd,rs,rt MUH rd,rs,rt MULU rd,rs,rt MUHU rd,rs,rt Purpose: Multiply Integers (with result to GPR) MUL: Multiply Words Signed, Low Word MUH: Multiply Words Signed, High Word MULU: Multiply Words Unsigned, Low Word MUHU: Multiply Words Unsigned, High Word Description: MUL: GPR[rd]  lo_word( multiply.signed( GPR[rs]  GPR[rt] ) ) MUH: GPR[rd]  hi_word( multiply.signed( GPR[rs]  GPR[rt] ) ) MULU: GPR[rd]  lo_word( multiply.unsigned( GPR[rs]  GPR[rt] ) ) MUHU: GPR[rd]  hi_word( multiply.unsigned( GPR[rs]  GPR[rt] ) ) MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 The Release 6 multiply instructions multiply the operands in GPR[rs] and GPR[rd], and place the specified high or low part of the result, of the same width, in GPR[rd]. MUL performs a signed 32-bit integer multiplication, and places the low 32 bits of the result in the destination regis- ter. MUH performs a signed 32-bit integer multiplication, and places the high 32 bits of the result in the destination regis- ter. MULU performs an unsigned 32-bit integer multiplication, and places the low 32 bits of the result in the destination register. MUHU performs an unsigned 32-bit integer multiplication, and places the high 32 bits of the result in the destination register. Restrictions: MUL behaves correctly even if its inputs are not sign extended 32-bit integers. Bits 32-63 of its inputs do not affect the result. MULU behaves correctly even if its inputs are not zero or sign extended 32-bit integers. Bits 32-63 of its inputs do not affect the result. The MIPS32® Instruction Set Manual, Revision 6.06 300 301 The MIPS32® Instruction Set Manual, Revision 6.06 MUL MUH MULU MUHU Multiply Integers (with result to GPR) Availability and Compatibility: These instructions are introduced by and required as of Release 6. Programming Notes: The low half of the integer multiplication result is identical for signed and unsigned. Nevertheless, there are distinct instructions MUL MULU. Implementations may choose to optimize a multiply that produces the low half followed by a multiply that produces the upper half. Programmers are recommended to use matching lower and upper half multiplications. The Release 6 MUL instruction has the same opcode mnemonic as the pre-Release 6 MUL instruction. The semantics of these instructions are almost identical: both produce the low 32-bits of the 3232=64 product; but the pre-Release 6 MUL is unpredictable if its inputs are not properly sign extended 32-bit values on a 64 bit machine, and is defined to render the HI and LO registers unpredictable, whereas the Release 6 version ignores bits 32-63 of the input, and there are no HI/LO registers in Release 6 to be affected. Operation: MUL, MUH: s1 signed_word(GPR[rs]) s2 signed_word(GPR[rt]) MULU, MUHU: s1 unsigned_word(GPR[rs]) s2 unsigned_word(GPR[rt]) product  s1  s2 MUL: GPR[rd]  MUH: GPR[rd]  MULU: GPR[rd]  MUHU: GPR[rd]  Exceptions: None /* product is twice the width of sources */ lo_word( product ) hi_word( product ) lo_word( product ) hi_word( product ) MUL.fmt IFloating Point Multiply 31 26 25 21 20 16 15 11 10 6 5 655556 COP1 010001 fmt ft fs fd MUL 000010 Format: MUL.fmt MUL.S fd, fs, ft MUL.D fd, fs, ft MUL.PS fd, fs, ft Purpose: Floating Point Multiply To multiply FP values. Description: FPR[fd]  FPR[fs] x FPR[ft] 0 MIPS32 MIPS32 MIPS64,MIPS32 Release 3, removed in Release 6 The value in FPR fs is multiplied by the value in FPR ft. The result is calculated to infinite precision, rounded accord- ing to the current rounding mode in FCSR, and placed into FPR fd. The operands and result are values in format fmt. MUL.PS multiplies the upper and lower halves of FPR fs and FPR ft independently, and ORs together any generated exceptional conditions. Restrictions: The fields fs, ft, and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPREDICTABLE. The operands must be values in format fmt; if they are not, the result is UNPREDICTABLE and the value of the operand FPRs becomes UNPREDICTABLE. The result of MUL.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: MUL.PS has been removed in Release 6. Operation: StoreFPR (fd, fmt, ValueFPR(fs, fmt) fmt ValueFPR(ft, fmt)) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Inexact, Unimplemented Operation, Invalid Operation, Overflow, Underflow The MIPS32® Instruction Set Manual, Revision 6.06 302 MULT Multiply Word 31 26 25 21 20 16 15 6 5 0 6 5 5 10 6 Format: MULT rs, rt MIPS32, removed in Release 6 Purpose: Multiply Word To multiply 32-bit signed integers. Description: (HI, LO)  GPR[rs] x GPR[rt] The 32-bit word value in GPR rt is multiplied by the 32-bit value in GPR rs, treating both operands as signed values, to produce a 64-bit result. The low-order 32-bit word of the result is placed into special register LO, and the high- order 32-bit word is placed into special register HI. No arithmetic exception occurs under any circumstances. Restrictions: None Availability and Compatibility: The MULT instruction has been removed in Release 6. It has been replaced by the Multiply Low (MUL) and Multiply High (MUH) instructions, whose output is written to a single GPR. Refer to the ‘MUL’ and ‘MUH’ instructions in this manual for more information. Operation: prod  GPR[rs]31..0 x GPR[rt]31..0 LO  prod31..0 HI  prod63..32 Exceptions: None Programming Notes: In some processors the integer multiply operation may proceed asynchronously and allow other CPU instructions to execute before it is complete. An attempt to read LO or HI before the results are written interlocks until the results are ready. Asynchronous execution does not affect the program result, but offers an opportunity for performance improvement by scheduling the multiply so that other instructions can execute in parallel. Programs that require overflow detection must check for it explicitly. Where the size of the operands are known, software should place the shorter operand in GPR rt. This may reduce the latency of the instruction on those processors which implement data-dependent instruction latencies. Implementation Note: SPECIAL 000000 rs rt 0 00 0000 0000 MULT 011000 303 The MIPS32® Instruction Set Manual, Revision 6.06 MULTU IMultiply Unsigned Word 31 26 25 21 20 16 15 6 5 0 6 5 5 10 6 Format: MULTU rs, rt MIPS32, removed in Release 6 Purpose: Multiply Unsigned Word To multiply 32-bit unsigned integers. Description: (HI, LO)  GPR[rs] x GPR[rt] The 32-bit word value in GPR rt is multiplied by the 32-bit value in GPR rs, treating both operands as unsigned val- ues, to produce a 64-bit result. The low-order 32-bit word of the result is placed into special register LO, and the high- order 32-bit word is placed into special register HI. No arithmetic exception occurs under any circumstances. Restrictions: None Availability and Compatibility: The MULTU instruction has been removed in Release 6. It has been replaced by the Multiply Low (MULU) and Mul- tiply High (MUHU) instructions, whose output is written to a single GPR. Refer to the ‘MULU’ and ‘MUHU’ instructions in this manual for more information. Operation: prod  (0 || GPR[rs]31..0) x (0 || GPR[rt]31..0) LO  prod31..0 HI  prod63..32 Exceptions: None Programming Notes: In some processors the integer multiply operation may proceed asynchronously and allow other CPU instructions to execute before it is complete. An attempt to read LO or HI before the results are written interlocks until the results are ready. Asynchronous execution does not affect the program result, but offers an opportunity for performance improvement by scheduling the multiply so that other instructions can execute in parallel. Programs that require overflow detection must check for it explicitly. Where the size of the operands are known, software should place the shorter operand in GPR rt. This may reduce the latency of the instruction on those processors which implement data-dependent instruction latencies. SPECIAL 000000 rs rt 0 00 0000 0000 MULTU 011001 The MIPS32® Instruction Set Manual, Revision 6.06 304 MULTU Multiply Unsigned Word 305 The MIPS32® Instruction Set Manual, Revision 6.06 NAL INo-op and Link 31 26 25 21 20 16 15 0 655 16 Format: NAL AssemblyIdiomMIPS32pre-Release6,MIPS32Release6 Purpose: No-op and Link Description:GPR[31] PC+8 NAL is an instruction used to read the PC. NAL was originally an alias for pre-Release 6 instruction BLTZAL. The condition is false, so the 16-bit target offset field is ignored, but the link register, GPR 31, is unconditionally written with the address of the instruction past the delay slot. Restrictions: NAL is considered to be a not-taken branch, with a delay slot, and may not be followed by instructions not allowed in delay slots. Nor is NAL allowed in a delay slot or forbidden slot. Availability and Compatibility: This is a deprecated instruction in Release 6. It is strongly recommended not to use this deprecated instructions because it will be removed from a future revision of the MIPS Architecture. The pre-Release 6 instruction BLTZAL when rs is not GPR[0], is removed in Release 6, and is required to signal a Reserved Instruction exception. Release 6 adds BLTZALC, the equivalent compact conditional branch and link, with no delay slot. This instruction, NAL, is introduced by and required as of Release 6, the mnemonic NAL becomes distinguished from the BLTZAL instruction removed in Release 6. The NAL instruction encoding, however, works on all imple- mentations, both pre-Release 6, where it was a special case of BLEZAL, and Release 6, where it is an instruction in its own right. NAL is provided only for compatibility with pre-Release 6 software. It is recommended that you use ADDIUPC to generate a PC-relative address. Exceptions: None Operation: GPR[31]  PC + 8 REGIMM 000001 0 00000 NAL 10000 offset The MIPS32® Instruction Set Manual, Revision 6.06 306 NEG.fmt Floating Point Negate 31 26 25 21 20 16 15 11 10 6 5 655556 COP1 010001 fmt 0 00000 fs fd NEG 000111 307 The MIPS32® Instruction Set Manual, Revision 6.06 Format: NEG.fmt NEG.S fd, fs NEG.D fd, fs NEG.PS fd, fs Purpose: Floating Point Negate To negate an FP value. Description: FPR[fd]  -FPR[fs] 0 MIPS32 MIPS32 MIPS32 Release 2, removed in Release 6 The value in FPR fs is negated and placed into FPR fd. The value is negated by changing the sign bit value. The oper- and and result are values in format fmt. NEG.PS negates the upper and lower halves of FPR fs independently, and ORs together any generated exceptional conditions. If FIRHas2008=0 or FCSRABS2008=0 then this operation is arithmetic. For this case, any NaN operand signals invalid operation. If FCSRABS2008=1 then this operation is non-arithmetic. For this case, both regular floating point numbers and NAN values are treated alike, only the sign bit is affected by this instruction. No IEEE 754 exception can be generated for this case, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPRE- DICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. The result of NEG.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: NEG.PS has been removed in Release 6. Operation: StoreFPR(fd, fmt, Negate(ValueFPR(fs, fmt))) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Unimplemented Operation, Invalid Operation NMADD.fmt IFloating Point Negative Multiply Add 31 26 25 21 20 16 15 11 10 6 5 3 2 0 6555533 COP1X 010011 fr ft fs fd NMADD 110 fmt Format: NMADD.fmt NMADD.S fd, fr, fs, ft NMADD.D fd, fr, fs, ft NMADD.PS fd, fr, fs, ft Purpose: Floating Point Negative Multiply Add To negate a combined multiply-then-add of FP values. MIPS32 Release 2, removed in Release 6 MIPS32 Release 2, removed in Release 6 MIPS32 Release 2, removed in Release 6 Description: FPR[fd]   ((FPR[fs] x FPR[ft])  FPR[fr]) The value in FPR fs is multiplied by the value in FPR ft to produce an intermediate product. The intermediate product is rounded according to the current rounding mode in FCSR. The value in FPR fr is added to the product. The result sum is calculated to infinite precision, rounded according to the current rounding mode in FCSR, negated by changing the sign bit, and placed into FPR fd. The operands and result are values in format fmt. The results and flags are as if separate floating-point multiply and add and negate instructions were executed. NMADD.PS applies the operation to the upper and lower halves of FPR fr, FPR fs, and FPR ft independently, and ORs together any generated exceptional conditions. The Cause bits are ORed into the Flag bits if no exception is taken. Restrictions: The fields fr, fs, ft, and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPREDICTABLE. The operands must be values in format fmt; if they are not, the result is UNPREDICTABLE and the value of the operand FPRs becomes UNPREDICTABLE. The result of NMADD.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: This instruction has been removed in Release 6. NMADD.S and NMADD.D: Required in all versions of MIPS64 since MIPS64 Release 1. Not available in MIPS32 Release 1. Required by MIPS32 Release 2 and subsequent versions of MIPS32. When required, these instructions are to be implemented if an FPU is present, either in a 32-bit or 64-bit FPU or in a 32-bit or 64-bit FP Register Mode (FIRF64=0 or 1, StatusFR=0 or 1). Operation: vfr  ValueFPR(fr, fmt) vfs  ValueFPR(fs, fmt) vft  ValueFPR(ft, fmt) StoreFPR(fd, fmt, (vfr fmt (vfs xfmt vft))) Exceptions: Coprocessor Unusable, Reserved Instruction The MIPS32® Instruction Set Manual, Revision 6.06 308 NMADD.fmt Floating Point Negative Multiply Add Floating Point Exceptions: Inexact, Unimplemented Operation, Invalid Operation, Overflow, Underflow 309 The MIPS32® Instruction Set Manual, Revision 6.06 NMSUB.fmt IFloating Point Negative Multiply Subtract 31 26 25 21 20 16 15 11 10 6 5 3 2 0 6555533 COP1X 010011 fr ft fs fd NMSUB 111 fmt Format: NMSUB.fmt NMSUB.S fd, fr, fs, ft NMSUB.D fd, fr, fs, ft NMSUB.PS fd, fr, fs, ft Purpose: Floating Point Negative Multiply Subtract To negate a combined multiply-then-subtract of FP values. MIPS32 Release 2, removed in Release 6 MIPS32 Release 2, removed in Release 6 MIPS32 Release 2, removed in Release 6 Description: FPR[fd]  ((FPR[fs] x FPR[ft])  FPR[fr]) The value in FPR fs is multiplied by the value in FPR ft to produce an intermediate product. The intermediate product is rounded according to the current rounding mode in FCSR. The value in FPR fr is subtracted from the product. The result is calculated to infinite precision, rounded according to the current rounding mode in FCSR, negated by changing the sign bit, and placed into FPR fd. The operands and result are values in format fmt. The results and flags are as if separate floating-point multiply and subtract and negate instructions were executed. NMSUB.PS applies the operation to the upper and lower halves of FPR fr, FPR fs, and FPR ft independently, and ORs together any generated exceptional conditions. The Cause bits are ORed into the Flag bits if no exception is taken. Restrictions: The fields fr, fs, ft, and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPREDICTABLE. The operands must be values in format fmt; if they are not, the result is UNPREDICTABLE and the value of the operand FPRs becomes UNPREDICTABLE. The result of NMSUB.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0 and not on a 32-bit FPU. Availability and Compatibility: This instruction has been removed in Release 6. NMSUB.S and NMSUB.D: Required in all versions of MIPS64 since MIPS64 Release 1. Not available in MIPS32 Release 1. Required in MIPS32 Release 2 and all subsequent versions of MIPS32. When required, these instructions are to be implemented if an FPU is present, either in a 32-bit or 64-bit FPU or in a 32-bit or 64-bit FP Register Mode (FIRF64=0 or 1, StatusFR=0 or 1). Operation: vfr  ValueFPR(fr, fmt) vfs  ValueFPR(fs, fmt) vft  ValueFPR(ft, fmt) StoreFPR(fd, fmt, ((vfs xfmt vft) fmt vfr)) Exceptions: Coprocessor Unusable, Reserved Instruction The MIPS32® Instruction Set Manual, Revision 6.06 310 NMSUB.fmt Floating Point Negative Multiply Subtract Floating Point Exceptions: Inexact, Unimplemented Operation, Invalid Operation, Overflow, Underflow 311 The MIPS32® Instruction Set Manual, Revision 6.06 NOP 31 INo Operation 0 26 25 21 20 16 15 11 10 6 5 655556 Format: NOP Purpose: No Operation To perform no operation. Description: Assembly Idiom SPECIAL 000000 0 00000 0 00000 0 00000 0 00000 SLL 000000 NOP is the assembly idiom used to denote no operation. The actual instruction is interpreted by the hardware as SLL r0, r0, 0. Restrictions: None Operations: None Exceptions: None Programming Notes: The zero instruction word, which represents SLL, r0, r0, 0, is the preferred NOP for software to use to fill branch and jump delay slots and to pad out alignment sequences. The MIPS32® Instruction Set Manual, Revision 6.06 312 NOR Not Or 31 26 25 21 20 16 15 11 10 6 5 655556 Format: NOR rd, rs, rt Purpose: Not Or To do a bitwise logical NOT OR. 0 MIPS32 SPECIAL 000000 rs rt rd 0 00000 NOR 100111 313 The MIPS32® Instruction Set Manual, Revision 6.06 Description: GPR[rd]  GPR[rs] nor GPR[rt] The contents of GPR rs are combined with the contents of GPR rt in a bitwise logical NOR operation. The result is placed into GPR rd. Restrictions: None Operation: GPR[rd]  GPR[rs] nor GPR[rt] Exceptions: None OR IOr 31 26 25 21 20 16 15 11 10 6 5 655556 Format: OR rd, rs, rt Purpose: Or To do a bitwise logical OR. 0 MIPS32 SPECIAL 000000 rs rt rd 0 00000 OR 100101 Description: GPR[rd]  GPR[rs] or GPR[rt] The contents of GPR rs are combined with the contents of GPR rt in a bitwise logical OR operation. The result is placed into GPR rd. Restrictions: None Operations: GPR[rd]  GPR[rs] or GPR[rt] Exceptions: None The MIPS32® Instruction Set Manual, Revision 6.06 314 ORI Or Immediate 31 26 25 21 20 16 15 655 16 Format: ORI rt, rs, immediate Purpose: Or Immediate To do a bitwise logical OR with a constant. 0 MIPS32 ORI 001101 rs rt immediate 315 The MIPS32® Instruction Set Manual, Revision 6.06 Description: GPR[rt]  GPR[rs] or immediate The 16-bit immediate is zero-extended to the left and combined with the contents of GPR rs in a bitwise logical OR operation. The result is placed into GPR rt. Restrictions: None Operations: GPR[rt]  GPR[rs] or zero_extend(immediate) Exceptions: None ORI IOr Immediate The MIPS32® Instruction Set Manual, Revision 6.06 316 PAUSE Wait for the LLBit to clear. 31 26 25 24 21 20 16 15 11 10 6 5 0 655556 Format: PAUSE MIPS32Release2/MTModule Purpose: Wait for the LLBit to clear. Description: Locks implemented using the LL/SC instructions are a common method of synchronization between threads of con- trol. A lock implementation does a load-linked instruction and checks the value returned to determine whether the software lock is set. If it is, the code branches back to retry the load-linked instruction, implementing an active busy- wait sequence. The PAUSE instruction is intended to be placed into the busy-wait sequence to block the instruction stream until such time as the load-linked instruction has a chance to succeed in obtaining the software lock. The PAUSE instruction is implementation-dependent, but it usually involves descheduling the instruction stream until the LLBit is zero. • In a single-threaded processor, this may be implemented as a short-term WAIT operation which resumes at the next instruction when the LLBit is zero or on some other external event such as an interrupt. • On a multi-threaded processor, this may be implemented as a short term YIELD operation which resumes at the next instruction when the LLBit is zero. In either case, it is assumed that the instruction stream which gives up the software lock does so via a write to the lock variable, which causes the processor to clear the LLBit as seen by this thread of execution. The encoding of the instruction is such that it is backward compatible with all previous implementations of the archi- tecture. The PAUSE instruction can therefore be placed into existing lock sequences and treated as a NOP by the pro- cessor, even if the processor does not implement the PAUSE instruction. Restrictions: Pre-Release 6: The operation of the processor is UNPREDICTABLE if a PAUSE instruction is executed placed in the delay slot of a branch or jump instruction. Release 6: Implementations are required to signal a Reserved Instruction exception if PAUSE is encountered in the delay slot or forbidden slot of a branch or jump instruction. SPECIAL 000000 0 00000 0 00000 0 00000 5 00101 SLL 000000 317 The MIPS32® Instruction Set Manual, Revision 6.06 Operations: if LLBit ≠ 0 then EPC  PC + 4 DescheduleInstructionStream() endif Exceptions: None Programming Notes: /* Resume at the following instruction */ The PAUSE instruction is intended to be inserted into the instruction stream after an LL instruction has set the LLBit and found the software lock set. The program may wait forever if a PAUSE instruction is executed and there is no possibility that the LLBit will ever be cleared. An example use of the PAUSE instruction is shown below: PAUSE IWait for the LLBit to clear. acquire_lock: ll t0, bnezc t0, addiu t0, sc t0, bnezc t0, sync 0(a0) /* Read software lock, set hardware lock */ acquire_lock_retry:/* Branch if software lock is taken; */ acquire_lock_retry: pause bc acquire_lock Critical region code release_lock: sync sw zero, 0(a0) 10: The MIPS32® Instruction Set Manual, Revision 6.06 318 t0, 1 0(a0) 10f /* Release 6 branch */ /* Set the software lock */ /* Try to store the software lock */ /* Branch if lock acquired successfully */ /* Wait for LLBIT to clear before retry */ /* and retry the operation; Release 6 branch */ /* Release software lock, clearing LLBIT */ /* for any PAUSEd waiters */ PLL.PS Pair Lower Lower 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: PLL.PS fd, fs, ft MIPS32 Release 2, removed in Release 6 Purpose: Pair Lower Lower To merge a pair of paired single values with realignment. Description: FPR[fd]  lower(FPR[fs]) || lower(FPR[ft]) A new paired-single value is formed by catenating the lower single of FPR fs (bits 31..0) and the lower single of FPR ft (bits 31..0). The move is non-arithmetic; it causes no IEEE 754 exceptions, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs, ft, and fd must specify FPRs valid for operands of type PS. If the fields are not valid, the result is UNPREDICTABLE. The result of this instruction is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: This instruction has been removed in Release 6. Operation: StoreFPR(fd, PS, ValueFPR(fs, PS)31..0 || ValueFPR(ft, PS)31..0) Exceptions: Coprocessor Unusable, Reserved Instruction COP1 010001 fmt 10110 ft fs fd PLL 101100 319 The MIPS32® Instruction Set Manual, Revision 6.06 PLU.PS IPair Lower Upper 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: PLU.PS fd, fs, ft MIPS32 Release 2, removed in Release 6 Purpose: Pair Lower Upper To merge a pair of paired single values with realignment. Description: FPR[fd]  lower(FPR[fs]) || upper(FPR[ft]) A new paired-single value is formed by catenating the lower single of FPR fs (bits 31..0) and the upper single of FPR ft (bits 63..32). The move is non-arithmetic; it causes no IEEE 754 exceptions, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs, ft, and fd must specify FPRs valid for operands of type PS. If the fields are not valid, the result is UNPREDICTABLE. The result of this instruction is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: This instruction has been removed in Release 6. Operation: StoreFPR(fd, PS, ValueFPR(fs, PS)31..0 || ValueFPR(ft, PS)63..32) Exceptions: Coprocessor Unusable, Reserved Instruction COP1 010001 fmt 10110 ft fs fd PLU 101101 The MIPS32® Instruction Set Manual, Revision 6.06 320 PREF Prefetch pre-Release 6 31 Release 6 26 25 21 20 16 15 655 16 PREF 110011 base hint offset 31 0 7 6 5 0 16 MIPS32 Description: prefetch_memory(GPR[base] + offset) PREF adds the signed offset to the contents of GPR base to form an effective byte address. The hint field supplies information about the way that the data is expected to be used. PREF enables the processor to take some action, typically causing data to be moved to or from the cache, to improve program performance. The action taken for a specific PREF instruction is both system and context dependent. Any action, including doing nothing, is permitted as long as it does not change architecturally visible state or alter the meaning of a program. Implementations are expected either to do nothing, or to take an action that increases the per- formance of the program. The PrepareForStore function is unique in that it may modify the architecturally visible state. PREF does not cause addressing-related exceptions, including TLB exceptions. If the address specified would cause an addressing exception, the exception condition is ignored and no data movement occurs.However even if no data is moved, some action that is not architecturally visible, such as writeback of a dirty cache line, can take place. It is implementation dependent whether a Bus Error or Cache Error exception is reported if such an error is detected as a byproduct of the action taken by the PREF instruction. PREF neither generates a memory operation nor modifies the state of a cache line for a location with an uncached memory access type, whether this type is specified by the address segment (e.g., kseg1), the programmed cacheability and coherency attribute of a segment (e.g., the use of the K0, KU, or K23 fields in the Config register), or the per- page cacheability and coherency attribute provided by the TLB. If PREF results in a memory operation, the memory access type and cacheability&coherency attribute used for the operation are determined by the memory access type and cacheability&coherency attribute of the effective address, just as it would be if the memory operation had been caused by a load or store to the effective address. For a cached location, the expected and useful action for the processor is to prefetch a block of data that includes the effective address. The size of the block and the level of the memory hierarchy it is fetched into are implementation specific. In coherent multiprocessor implementations, if the effective address uses a coherent Cacheability and Coherency Attribute (CCA), then the instruction causes a coherent memory transaction to occur. This means a prefetch issued on one processor can cause data to be evicted from the cache in another processor. The PREF instruction and the memory transactions which are sourced by the PREF instruction, such as cache refill or cache writeback, obey the ordering and completion rules of the SYNC instruction. 26 25 21 20 16 15 6559 Format: PREF hint,offset(base) Purpose: Prefetch To move data between memory and cache. SPECIAL3 011111 base hint offset 0 PREF 110101 321 The MIPS32® Instruction Set Manual, Revision 6.06 PREF IPrefetch Table 5.2 Values of hint Field for PREF Instruction Value Name Data Use and Desired Prefetch Action 0 load Use: Prefetched data is expected to be read (not modified). Action: Fetch data as if for a load. 1 store Use: Prefetched data is expected to be stored or modified. Action: Fetch data as if for a store. 2 L1 LRU hint Pre-Release 6: Reserved for Architecture. Release 6: Implementation dependent. This hint code marks the line as LRU in the L1 cache and thus preferred for next eviction. Implementations can choose to writeback and/or invalidate as long as no architectural state is modified. 3 Reserved for Implementation Pre-Release 6: Reserved for Architecture. Release 6: Available for implementation-dependent use. 4 load_streamed Use: Prefetched data is expected to be read (not modified) but not reused extensively; it “streams” through cache. Action: Fetch data as if for a load and place it in the cache so that it does not displace data prefetched as “retained.” 5 store_streamed Use: Prefetched data is expected to be stored or modified but not reused exten- sively; it “streams” through cache. Action: Fetch data as if for a store and place it in the cache so that it does not displace data prefetched as “retained.” 6 load_retained Use: Prefetched data is expected to be read (not modified) and reused exten- sively; it should be “retained” in the cache. Action: Fetch data as if for a load and place it in the cache so that it is not dis- placed by data prefetched as “streamed.” 7 store_retained Use: Prefetched data is expected to be stored or modified and reused exten- sively; it should be “retained” in the cache. Action: Fetch data as if for a store and place it in the cache so that it is not dis- placed by data prefetched as “streamed.” 8-15 L2 operation Pre-Release 6: Reserved for Architecture. Release 6: In the Release 6 architecture, hint codes 8 - 15 are treated the same as hint codes 0 - 7 respectively, but operate on the L2 cache. 16-23 L3 operation Pre-Release 6: Reserved for Architecture. Release 6: In the Release 6 architecture, hint codes 16 - 23 are treated the same as hint codes 0 - 7 respectively, but operate on the L3 cache. 24 Reserved for Architecture Pre-Release 6: Unassigned by the Architecture - available for implementation- dependent use. Release 6: This hint code is not implemented in the Release 6 architecture and generates a Reserved Instruction exception (RI). The MIPS32® Instruction Set Manual, Revision 6.06 322 PREF Prefetch Table 5.2 Values of hint Field for PREF Instruction (Continued) Value Name Data Use and Desired Prefetch Action 25 writeback_invalidate (also known as “nudge”) Reserved for Architecture in Release 6 Pre-Release 6: Use—Data is no longer expected to be used. Action—For a writeback cache, schedule a writeback of any dirty data. At the completion of the writeback, mark the state of any cache lines written back as invalid. If the cache line is not dirty, it is implementation dependent whether the state of the cache line is marked invalid or left unchanged. If the cache line is locked, no action is taken. Release 6: This hint code is not implemented in the Release 6 architecture and generates a Reserved Instruction exception (RI). 26-29 Reserved for Architecture Pre-Release 6: Unassigned by the Architecture—available for implementa- tion-dependent use. Release 6: These hints are not implemented in the Release 6 architecture and generate a Reserved Instruction exception (RI). 30 PrepareForStore Reserved for Architecture in Release 6 Pre-Release 6: Use—Prepare the cache for writing an entire line, without the overhead involved in filling the line from memory. Action—If the reference hits in the cache, no action is taken. If the reference misses in the cache, a line is selected for replacement, any valid and dirty vic- tim is written back to memory, the entire line is filled with zero data, and the state of the line is marked as valid and dirty. Programming Note: Because the cache line is filled with zero data on a cache miss, software must not assume that this action, in and of itself, can be used as a fast bzero-type function. Release 6: This hint is not implemented in the Release 6 architecture and gen- erates a Reserved Instruction exception (RI). 31 Reserved for Architecture Pre-Release 6: Unassigned by the Architecture—available for implementa- tion-dependent use. Release 6: This hint is not implemented in the Release 6 architecture and gen- erates a Reserved Instruction exception (RI). 323 The MIPS32® Instruction Set Manual, Revision 6.06 Restrictions: None This instruction does not produce an exception for a misaligned memory address, since it has no memory access size. Availability and Compatibility: This instruction has been recoded for Release 6. Operation: vAddr  GPR[base]  sign_extend(offset) (pAddr, CCA)  AddressTranslation(vAddr, DATA, LOAD) Prefetch(CCA, pAddr, vAddr, DATA, hint) Exceptions: Bus Error, Cache Error Prefetch does not take any TLB-related or address-related exceptions under any circumstances. Programming Notes: In the Release 6 architecture, hint codes 2:3, 10:11, 18:19 behave as a NOP if not implemented. Hint codes 24:31 are PREF IPrefetch As shown in the instruction drawing above, Release 6 implements a 9-bit offset, whereas all release levels lower than Release 6 of the MIPS architecture implement a 16-bit offset. Prefetch cannot move data to or from a mapped location unless the translation for that location is present in the TLB. Locations in memory pages that have not been accessed recently may not have translations in the TLB, so prefetch may not be effective for such locations. Prefetch does not cause addressing exceptions. A prefetch may be used using an address pointer before the validity of the pointer is determined without worrying about an addressing exception. It is implementation dependent whether a Bus Error or Cache Error exception is reported if such an error is detected as a byproduct of the action taken by the PREF instruction. Typically, this only occurs in systems which have high- reliability requirements. Prefetch operations have no effect on cache lines that were previously locked with the CACHE instruction. Hint field encodings whose function is described as “streamed” or “retained” convey usage intent from software to hardware. Software should not assume that hardware will always prefetch data in an optimal way. If data is to be truly retained, software should use the Cache instruction to lock data into the cache. not implemented (treated as reserved) and always signal a Reserved Instruction exception (RI). The MIPS32® Instruction Set Manual, Revision 6.06 324 PREFE Prefetch EVA 31 26 25 21 20 16 15 7 6 5 0 655916 Format: PREFE hint,offset(base) MIPS32 Purpose: Prefetch EVA To move data between user mode virtual address space memory and cache while operating in kernel mode. Description: prefetch_memory(GPR[base] + offset) PREFE adds the 9-bit signed offset to the contents of GPR base to form an effective byte address. The hint field sup- plies information about the way that the data is expected to be used. PREFE enables the processor to take some action, causing data to be moved to or from the cache, to improve program performance. The action taken for a specific PREFE instruction is both system and context dependent. Any action, including doing nothing, is permitted as long as it does not change architecturally visible state or alter the meaning of a program. Implementations are expected either to do nothing, or to take an action that increases the performance of the program. The PrepareForStore function is unique in that it may modify the architecturally visible state. PREFE does not cause addressing-related exceptions, including TLB exceptions. If the address specified would cause an addressing exception, the exception condition is ignored and no data movement occurs.However even if no data is moved, some action that is not architecturally visible, such as writeback of a dirty cache line, can take place. It is implementation dependent whether a Bus Error or Cache Error exception is reported if such an error is detected as a byproduct of the action taken by the PREFE instruction. PREFE neither generates a memory operation nor modifies the state of a cache line for a location with an uncached memory access type, whether this type is specified by the address segment (for example, kseg1), the programmed cacheability and coherency attribute of a segment (for example, the use of the K0, KU, or K23 fields in the Config register), or the per-page cacheability and coherency attribute provided by the TLB. If PREFE results in a memory operation, the memory access type and cacheability & coherency attribute used for the operation are determined by the memory access type and cacheability & coherency attribute of the effective address, just as it would be if the memory operation had been caused by a load or store to the effective address. For a cached location, the expected and useful action for the processor is to prefetch a block of data that includes the effective address. The size of the block and the level of the memory hierarchy it is fetched into are implementation specific. In coherent multiprocessor implementations, if the effective address uses a coherent Cacheability and Coherency Attribute (CCA), then the instruction causes a coherent memory transaction to occur. This means a prefetch issued on one processor can cause data to be evicted from the cache in another processor. The PREFE instruction and the memory transactions which are sourced by the PREFE instruction, such as cache refill or cache writeback, obey the ordering and completion rules of the SYNC instruction. The PREFE instruction functions in exactly the same fashion as the PREF instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment configured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Enhanced Virtual Addressing section for additional information. Implementation of this instruction is specified by the Config5EVA field being set to one. SPECIAL3 011111 base hint offset 0 PREFE 100011 325 The MIPS32® Instruction Set Manual, Revision 6.06 PREFE IPrefetch EVA Table 5.3 Values of hint Field for PREFE Instruction Value Name Data Use and Desired Prefetch Action 0 load Use: Prefetched data is expected to be read (not modified). Action: Fetch data as if for a load. 1 store Use: Prefetched data is expected to be stored or modified. Action: Fetch data as if for a store. 2 L1 LRU hint Pre-Release 6: Reserved for Architecture. Release 6: Implementation dependent. This hint code marks the line as LRU in the L1 cache and thus preferred for next eviction. Implementations can choose to writeback and/or invalidate as long as no architectural state is modified. 3 Reserved for Implementation Pre-Release 6: Reserved for Architecture. Release 6: Available for implementation-dependent use. 4 load_streamed Use: Prefetched data is expected to be read (not modified) but not reused extensively; it “streams” through cache. Action: Fetch data as if for a load and place it in the cache so that it does not displace data prefetched as “retained.” 5 store_streamed Use: Prefetched data is expected to be stored or modified but not reused exten- sively; it “streams” through cache. Action: Fetch data as if for a store and place it in the cache so that it does not displace data prefetched as “retained.” 6 load_retained Use: Prefetched data is expected to be read (not modified) and reused exten- sively; it should be “retained” in the cache. Action: Fetch data as if for a load and place it in the cache so that it is not dis- placed by data prefetched as “streamed.” 7 store_retained Use: Prefetched data is expected to be stored or modified and reused exten- sively; it should be “retained” in the cache. Action: Fetch data as if for a store and place it in the cache so that it is not dis- placed by data prefetched as “streamed.” 8-15 L2 operation Pre-Release 6: Reserved for Architecture. Release 6: Hint codes 8 - 15 are treated the same as hint codes 0 - 7 respec- tively, but operate on the L2 cache. 16-23 L3 operation Pre-Release 6: Reserved for Architecture. Release 6: Hint codes 16 - 23 are treated the same as hint codes 0 - 7 respec- tively, but operate on the L3 cache. 24 Reserved for Architecture Pre-Release 6: Unassigned by the Architecture - available for implementation- dependent use. Release 6: This hint code is not implemented in the Release 6 architecture and generates a Reserved Instruction exception (RI). The MIPS32® Instruction Set Manual, Revision 6.06 326 PREFE Prefetch EVA Table 5.3 Values of hint Field for PREFE Instruction (Continued) Value Name Data Use and Desired Prefetch Action 25 writeback_invalidate (also known as “nudge”) Reserved for Architecture in Release 6 Pre-Release 6: Use—Data is no longer expected to be used. Action—For a writeback cache, schedule a writeback of any dirty data. At the completion of the writeback, mark the state of any cache lines written back as invalid. If the cache line is not dirty, it is implementation dependent whether the state of the cache line is marked invalid or left unchanged. If the cache line is locked, no action is taken. Release 6: This hint code is not implemented in the Release 6 architecture and generates a Reserved Instruction exception (RI). 26-29 Reserved for Architecture Pre-Release 6: Unassigned by the Architecture - available for implementation- dependent use. Release 6: These hint codes are not implemented in the Release 6 architecture and generate a Reserved Instruction exception (RI). 30 PrepareForStore Reserved for Architecture in Release 6 Pre-Release 6: Use—Prepare the cache for writing an entire line, without the overhead involved in filling the line from memory. Action—If the reference hits in the cache, no action is taken. If the reference misses in the cache, a line is selected for replacement, any valid and dirty vic- tim is written back to memory, the entire line is filled with zero data, and the state of the line is marked as valid and dirty. Programming Note: Because the cache line is filled with zero data on a cache miss, software must not assume that this action, in and of itself, can be used as a fast bzero-type function. Release 6: This hint code is not implemented in the Release 6 architecture and generates a Reserved Instruction exception (RI). 31 Reserved for Architecture Pre-Release 6: Unassigned by the Architecture - available for implementation- dependent use. Release 6: This hint code is not implemented in the Release 6 architecture and generates a Reserved Instruction exception (RI). 327 The MIPS32® Instruction Set Manual, Revision 6.06 Restrictions: Only usable when access to Coprocessor0 is enabled and when accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode. This instruction does not produce an exception for a misaligned memory address, since it has no memory access size. Operation: vAddr  GGPR[base]  sign_extend(offset) (pAddr, CCA)  AddressTranslation(vAddr, DATA, LOAD) Prefetch(CCA, pAddr, vAddr, DATA, hint) Exceptions: Bus Error, Cache Error, Address Error, Reserved Instruction, Coprocessor Usable Prefetch does not take any TLB-related or address-related exceptions under any circumstances. Programming Notes: In the Release 6 architecture, hint codes 0:23 behave as a NOP and never signal a Reserved Instruction exception (RI). Hint codes 24:31 are not implemented (treated as reserved) and always signal a Reserved Instruction exception (RI). PREFE IPrefetch EVA Prefetch cannot move data to or from a mapped location unless the translation for that location is present in the TLB. Locations in memory pages that have not been accessed recently may not have translations in the TLB, so prefetch may not be effective for such locations. Prefetch does not cause addressing exceptions. A prefetch may be used using an address pointer before the validity of the pointer is determined without worrying about an addressing exception. It is implementation dependent whether a Bus Error or Cache Error exception is reported if such an error is detected as a byproduct of the action taken by the PREFE instruction. Typically, this only occurs in systems which have high- reliability requirements. Prefetch operations have no effect on cache lines that were previously locked with the CACHE instruction. Hint field encodings whose function is described as “streamed” or “retained” convey usage intent from software to hardware. Software should not assume that hardware will always prefetch data in an optimal way. If data is to be truly retained, software should use the Cache instruction to lock data into the cache. The MIPS32® Instruction Set Manual, Revision 6.06 328 PREFX Prefetch Indexed 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: PREFX hint, index(base) MIPS64, MIPS32 Release 2, removed in Release 6 Purpose: Prefetch Indexed To move data between memory and cache. Description: prefetch_memory[GPR[base]  GPR[index]] PREFX adds the contents of GPR index to the contents of GPR base to form an effective byte address. The hint field supplies information about the way the data is expected to be used. The only functional difference between the PREF and PREFX instructions is the addressing mode implemented by the two. Refer to the PREF instruction for all other details, including the encoding of the hint field. Restrictions: Availability and Compatibility: Required in all versions of MIPS64 since MIPS64 Release 1. Not available in MIPS32 Release 1. Required by MIPS32 Release 2 and subsequent versions of MIPS32. When required, required whenever FPU is present, whether a 32-bit or 64-bit FPU, whether in 32-bit or 64-bit FP Register Mode (FIRF64=0 or 1, StatusFR=0 or 1). This instruction has been removed in Release 6. Operation: vAddr  GPR[base]  GPR[index] (pAddr, CCA)  AddressTranslation(vAddr, DATA, LOAD) Prefetch(CCA, pAddr, vAddr, DATA, hint) Exceptions: Coprocessor Unusable, Reserved Instruction, Bus Error, Cache Error Programming Notes: The PREFX instruction is only available on processors that implement floating point and should never by generated by compilers in situations other than those in which the corresponding load and store indexed floating point instruc- tions are generated. Refer to the corresponding section in the PREF instruction description. COP1X 010011 base index hint 0 00000 PREFX 001111 329 The MIPS32® Instruction Set Manual, Revision 6.06 PUL.PS IPair Upper Lower 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: PUL.PS fd, fs, ft MIPS64, MIPS32 Release 2, removed in Release 6 Purpose: Pair Upper Lower To merge a pair of paired single values with realignment. Description: FPR[fd]  upper(FPR[fs]) || lower(FPR[ft]) A new paired-single value is formed by catenating the upper single of FPR fs (bits 63..32) and the lower single of FPR ft (bits 31..0). The move is non-arithmetic; it causes no IEEE 754 exceptions, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs, ft, and fd must specify FPRs valid for operands of type PS. If the fields are not valid, the result is UNPREDICTABLE. The result of this instruction is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: This instruction has been removed in Release 6. Operation: StoreFPR(fd, PS, ValueFPR(fs, PS)63..32 || ValueFPR(ft, PS)31..0) Exceptions: Coprocessor Unusable, Reserved Instruction COP1 010001 fmt 10110 ft fs fd PUL 101110 The MIPS32® Instruction Set Manual, Revision 6.06 330 PUU.PS Pair Upper Upper 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: PUU.PS fd, fs, ft MIPS64,MIPS32 Release 2,, removed in Release 6 Purpose: Pair Upper Upper To merge a pair of paired single values with realignment. Description: FPR[fd]  upper(FPR[fs]) || upper(FPR[ft]) A new paired-single value is formed by catenating the upper single of FPR fs (bits 63..32) and the upper single of FPR ft (bits 63..32). The move is non-arithmetic; it causes no IEEE 754 exceptions, and the FCSRCause and FCSRFlags fields are not modified. Restrictions: The fields fs, ft, and fd must specify FPRs valid for operands of type PS. If the fields are not valid, the result is UNPREDICTABLE. The result of this instruction is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Availability and Compatibility: This instruction has been removed in Release 6. Operation: StoreFPR(fd, PS, ValueFPR(fs, PS)63..32 || ValueFPR(ft, PS)63..32) Exceptions: Coprocessor Unusable, Reserved Instruction COP1 010001 fmt 10110 ft fs fd PUU 101111 331 The MIPS32® Instruction Set Manual, Revision 6.06 RDHWR IRead Hardware Register 31 26 25 21 20 16 15 11 10 6 5 0 6555236 Format: RDHWR rt,rd,sel MIPS32 Release 2 Purpose: Read Hardware Register To move the contents of a hardware register to a general purpose register (GPR) if that operation is enabled by privi- leged software. The purpose of this instruction is to give user mode access to specific information that is otherwise only visible in kernel mode. In Release 6, a sel field has been added to allow a register with multiple instances to be read selectively. Specifically it is used for PerfCtr. Description: GPR[rt]  HWR[rd]; GPR[rt]  HWR[rd, sel] If access is allowed to the specified hardware register, the contents of the register specified by rd (optionally sel in Release 6) is loaded into general register rt. Access control for each register is selected by the bits in the coprocessor 0 HWREna register. The available hardware registers, and the encoding of the rd field for each, are shown in Table 5.4. Table 5.4 RDHWR Register Numbers SPECIAL3 011111 0 00000 rt rd 0 00 sel RDHWR 111011 Register Number (rd Value) Mnemonic Description 0 CPUNum Number of the CPU on which the program is currently running. This register pro- vides read access to the coprocessor 0 EBaseCPUNum field. 1 SYNCI_Step Address step size to be used with the SYNCI instruction, or zero if no caches need be synchronized. See that instruction’s description for the use of this value. 2 CC High-resolution cycle counter. This register provides read access to the coprocessor 0 Count Register. 3 CCRes Resolution of the CC register. This value denotes the number of cycles between update of the register. For example: CCRes Value Meaning 1 CC register increments every CPU cycle 2 CC register increments every second CPU cycle 3 CC register increments every third CPU cycle etc. 4 PerfCtr Performance Counter Pair. Even sel selects the Control register, while odd sel selects the Counter register in the pair. The value of sel corresponds to the value of sel used by MFC0 to read the CP0 register. The MIPS32® Instruction Set Manual, Revision 6.06 332 RDHWR Read Hardware Register Table 5.4 RDHWR Register Numbers Register Number (rd Value) Mnemonic Description 5 XNP Indicates support for the Release 6 Paired LL/SC family of instructions. If set to 1, the LL/SC family of instructions is not present, otherwise, it is present in the imple- mentation. In absence of hardware support for double-width or extended atomics, user software may emulate the instruction’s behavior through other means. See Config5XNP. 6-28 These registers numbers are reserved for future architecture use. Access results in a Reserved Instruction Exception. 29 ULR User Local Register. This register provides read access to the coprocessor 0 UserLocal register, if it is implemented. In some operating environments, the UserLocal register is a pointer to a thread-specific storage block. 30-31 These register numbers are reserved for implementation-dependent use. If they are not implemented, access results in a Reserved Instruction Exception. 333 The MIPS32® Instruction Set Manual, Revision 6.06 Restrictions: In implementations of Release 1 of the Architecture, this instruction resulted in a Reserved Instruction Exception. Access to the specified hardware register is enabled if Coprocessor 0 is enabled, or if the corresponding bit is set in the HWREna register. If access is not allowed or the register is not implemented, a Reserved Instruction Exception is signaled. In Release 6, when the 3-bit sel is undefined for use with a specific register number, then a Reserved Instruction Exception is signaled. Availability and Compatibility: This instructions has been recoded for Release 6. The instruction supports a sel field in Release 6. Operation: if ((rs!=4) and (sel==0)) case rd 0: temp  EBaseCPUNum 1: temp  SYNCI_StepSize() 2: temp  Count 3: temp  CountResolution() if (>=2) // #5 – Release 6 5: temp  Config5XNPendif
29: temp  UserLocal
endif
30: temp  Implementation-Dependent-Value
31: temp  Implementation-Dependent-Value otherwise: SignalException(ReservedInstruction)
endcase
elseif ((rs==4) and (>=2) and (sel==defined)// #4 – Release 6
temp  PerfCtr[sel] else
endif GPR[rt]  temp

RDHWR IRead Hardware Register Exceptions:
Reserved Instruction
For a register that does not require sel, the compiler must support an assembly syntax without sel that is ‘RDHWR rt, rd’. Another valid syntax is for sel to be 0 to map to pre-Release 6 register numbers which do not require use of sel that is, ‘RDHWR rt, rd, 0’.
The MIPS32® Instruction Set Manual, Revision 6.06 334

RDPGPR Read GPR from Previous Shadow Set
31
26 25 21 20 16 15 11 10
6 5 5 5 11
Format: RDPGPR rd, rt
Purpose: Read GPR from Previous Shadow Set
To move the contents of a GPR from the previous shadow set to a current GPR.
0
MIPS32 Release 2
COP0 0100 00
RDPGPR 01 010
rt
rd
0
000 0000 0000
335
The MIPS32® Instruction Set Manual, Revision 6.06
Description: GPR[rd]  SGPR[SRSCtlPSS, rt]
The contents of the shadow GPR register specified by SRSCtlPSS (signifying the previous shadow set number) and rt
(specifying the register number within that set) is moved to the current GPR rd. Restrictions:
In implementations prior to Release 2 of the Architecture, this instruction resulted in a Reserved Instruction excep- tion.
Operation:
GPR[rd]  SGPR[SRSCtlPSS, rt]
Exceptions:
Coprocessor Unusable Reserved Instruction

RECIP.fmt IReciprocal Approximation
31
26 25 21 20 16 15 11 10 6 5 0
655556
COP1 010001
fmt
0 00000
fs
fd
RECIP 010101
Format: RECIP.fmt RECIP.S fd, fs
RECIP.D fd, fs
Purpose: Reciprocal Approximation
To approximate the reciprocal of an FP value (quickly).
MIPS64,MIPS32 Release 2 MIPS64,MIPS32 Release 2
Description: FPR[fd]  1.0 / FPR[fs]
The reciprocal of the value in FPR fs is approximated and placed into FPR fd. The operand and result are values in
format fmt.
The numeric accuracy of this operation is implementation dependent. It does not meet the accuracy specified by the IEEE 754 Floating Point standard. The computed result differs from the both the exact result and the IEEE-mandated representation of the exact result by no more than one unit in the least-significant place (ULP).
It is implementation dependent whether the result is affected by the current rounding mode in FCSR. Restrictions:
The fields fs and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPRE- DICTABLE.
The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE.
Availability and Compatibility:
RECIP.S and RECIP.D: Required in all versions of MIPS64 since MIPS64 Release 1. Not available in MIPS32 Release 1. Required in MIPS32 Release 2 and all subsequent versions of MIPS32. When required, required whenever FPU is present, whether a 32-bit or 64-bit FPU, whether in 32-bit or 64-bit FP Register Mode (FIRF64=0 or 1, StatusFR=0 or 1).
Operation:
StoreFPR(fd, fmt, 1.0 / valueFPR(fs, fmt))
Exceptions:
Coprocessor Unusable, Reserved Instruction
Floating Point Exceptions:
Inexact, Division-by-zero, Unimplemented Op, Invalid Op, Overflow, Underflow
The MIPS32® Instruction Set Manual, Revision 6.06 336

RINT.fmt Floating-Point Round to Integral
31
26 25 21 20 16 15 11 10 6 5 0
655556
COP1 010001
fmt
00000
fs
fd
RINT 011010
337
The MIPS32® Instruction Set Manual, Revision 6.06
Format: RINT.fmt RINT.S fd,fs
RINT.D fd,fs
Purpose: Floating-Point Round to Integral
Scalar floating-point round to integral floating point value.
Description:FPR[fd] round_int(FPR[fs])
MIPS32Release6 MIPS32 Release 6 MIPS32 Release 6
The scalar floating-point value in the register fs is rounded to an integral valued floating-point number in the same format based on the rounding mode bits RM in the FPU Control and Status Register FCSR. The result is written to fd.
The operands and results are values in floating-point data format fmt.
The RINT.fmt instruction corresponds to the roundToIntegralExact operation in the IEEE Standard for Floating-
Point Arithmetic 754TM-2008. The Inexact exception is signaled if the result does not have the same numerical value as the input operand.
The floating point scalar instruction RINT.fmt corresponds to the MSA vector instruction FRINT.df. I.e. RINT.S cor- responds to FRINT.W, and RINT.D corresponds to FRINT.D.
Restrictions:
Data-dependent exceptions are possible as specified by the IEEE Standard for Floating-Point Arithmetic 754TM- 2008.
Availability and Compatibility:
This instruction is introduced by and required as of Release 6.
Operation:
RINT fmt:
if not IsCoprocessorEnabled(1)
then SignalException(CoprocessorUnusable, 1) endif
if not IsFloatingPointImplemented(fmt))
then SignalException(ReservedInstruction) endif
fin  ValueFPR(fs,fmt)
ftmp RoundIntFP(fin, fmt)
if( fin  ftmp ) SignalFPException(InExact) StoreFPR (fd, fmt, ftmp )
function RoundIntFP(tt, n)
/* Round to integer operation, using rounding mode FCSR.RM*/
endfunction RoundIntFP
Exceptions:
Coprocessor Unusable, Reserved Instruction

RINT.fmt IFloating-Point Round to Integral Floating Point Exceptions:
Unimplemented Operation, Invalid Operation, Inexact, Overflow, Underflow
The MIPS32® Instruction Set Manual, Revision 6.06 338

ROTR Rotate Word Right
31
26 25 22 21 20 16 15 11 10 6 5 0
6415556
Format: ROTR rd, rt, sa SmartMIPS Crypto, MIPS32 Release 2 Purpose: Rotate Word Right
To execute a logical right-rotate of a word by a fixed number of bits.
Description: GPR[rd]  GPR[rt] (right) sa
The contents of the low-order 32-bit word of GPR rt are rotated right; the word result is placed in GPR rd. The bit-
rotate amount is specified by sa.
Restrictions:
Operation:
if ((ArchitectureRevision()  2) and (Config3SM = 0)) then UNPREDICTABLE
endif
ssa
temp  GPR[rt]s-1..0 || GPR[rt]31..s GPR[rd]  temp
Exceptions:
Reserved Instruction
SPECIAL 000000
0000
R 1
rt
rd
sa
SRL 000010
339
The MIPS32® Instruction Set Manual, Revision 6.06

ROTRV IRotate Word Right Variable
31
26 25 21 20 16 15 11 10 7 6 5 0
6555416
Format: ROTRV rd, rt, rs SmartMIPS Crypto, MIPS32 Release 2 Purpose: Rotate Word Right Variable
To execute a logical right-rotate of a word by a variable number of bits.
Description: GPR[rd]  GPR[rt] (right) GPR[rs]
The contents of the low-order 32-bit word of GPR rt are rotated right; the word result is placed in GPR rd. The bit-
rotate amount is specified by the low-order 5 bits of GPR rs.
Restrictions:
Operation:
if ((ArchitectureRevision()  2) and (Config3SM = 0)) then UNPREDICTABLE
endif
s  GPR[rs]4..0
temp  GPR[rt]s-1..0 || GPR[rt]31..s GPR[rd]  temp
Exceptions:
Reserved Instruction
SPECIAL 000000
rs
rt
rd
0000
R 1
SRLV 000110
The MIPS32® Instruction Set Manual, Revision 6.06 340

ROUND.L.fmt Floating Point Round to Long Fixed Point
31
26 25 21 20 16 15 11 10 6 5 0
655556
COP1 010001
fmt
0 00000
fs
fd
ROUND.L 001000
341
The MIPS32® Instruction Set Manual, Revision 6.06
Format: ROUND.L.fmt ROUND.L.S fd, fs
ROUND.L.D fd, fs
Purpose: Floating Point Round to Long Fixed Point
To convert an FP value to 64-bit fixed point, rounding to nearest.
MIPS64,MIPS32 Release 2 MIPS64,MIPS32 Release 2
Description: FPR[fd]  convert_and_round(FPR[fs])
The value in FPR fs, in format fmt, is converted to a value in 64-bit long fixed point format and rounded to nearest/
even (rounding mode 0). The result is placed in FPR fd.
When the source value is Infinity, NaN, or rounds to an integer outside the range -263 to 263-1, the result cannot be represented correctly and an IEEE Invalid Operation condition exists. The Invalid Operation flag is set in the FCSR. If the Invalid Operation Enable bit is set in the FCSR, no result is written to fd and an Invalid Operation exception is taken immediately. Otherwise, a default result is written to fd. On cores with FCSRNAN2008=0, the default result is
263–1. On cores with FCSRNAN2008=1, the default result is:
• 0 when the input value is NaN
• 263–1 when the input value is + or rounds to a number larger than 263–1
• -263–1 when the input value is – or rounds to a number smaller than -263–1
Restrictions:
The fields fs and fd must specify valid FPRs: fs for type fmt and fd for long fixed point. If the fields are not valid, the result is UNPREDICTABLE.
The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE.
The result of this instruction is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. It is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU.
Operation:
StoreFPR(fd, L, ConvertFmt(ValueFPR(fs, fmt), fmt, L))
Exceptions:
Coprocessor Unusable, Reserved Instruction
Floating Point Exceptions:
Inexact, Unimplemented Operation, Invalid Operation

ROUND.W.fmt IFloating Point Round to Word Fixed Point
31
26 25 21 20 16 15 11 10 6 5
655556
Format: ROUND.W.fmt ROUND.W.S fd, fs
ROUND.W.D fd, fs
Purpose: Floating Point Round to Word Fixed Point
To convert an FP value to 32-bit fixed point, rounding to nearest.
0
MIPS32 MIPS32
COP1 010001
fmt
0 00000
fs
fd
ROUND.W 001100
Description: FPR[fd]  convert_and_round(FPR[fs])
The value in FPR fs, in format fmt, is converted to a value in 32-bit word fixed point format rounding to nearest/even
(rounding mode 0). The result is placed in FPR fd.
When the source value is Infinity, NaN, or rounds to an integer outside the range -231 to 231-1, the result cannot be represented correctly and an IEEE Invalid Operation condition exists. The Invalid Operation flag is set in the FCSR. If the Invalid Operation Enable bit is set in the FCSR, no result is written to fd and an Invalid Operation exception is taken immediately. Otherwise, a default result is written to fd. On cores with FCSRNAN2008=0, the default result is
231–1. On cores with FCSRNAN2008=1, the default result is:
• 0 when the input value is NaN
• 231–1 when the input value is + or rounds to a number larger than 231–1
• -231–1 when the input value is – or rounds to a number smaller than -231–1
Restrictions:
The fields fs and fd must specify valid FPRs: fs for type fmt and fd for word fixed point. If the fields are not valid, the result is UNPREDICTABLE.
The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE.
Operation:
StoreFPR(fd, W, ConvertFmt(ValueFPR(fs, fmt), fmt, W))
Exceptions:
Coprocessor Unusable, Reserved Instruction
Floating Point Exceptions:
Inexact, Unimplemented Operation, Invalid Operation
The MIPS32® Instruction Set Manual, Revision 6.06 342

RSQRT.fmt Reciprocal Square Root Approximation
31
26 25 21 20 16 15 11 10 6 5 0
655556
COP1 010001
fmt
0 00000
fs
fd
RSQRT 010110
343
The MIPS32® Instruction Set Manual, Revision 6.06
Format: RSQRT.fmt RSQRT.S fd, fs
RSQRT.D fd, fs
Purpose: Reciprocal Square Root Approximation
To approximate the reciprocal of the square root of an FP value (quickly).
MIPS64,MIPS32 Release 2 MIPS64,MIPS32 Release 2
Description: FPR[fd]  1.0 / sqrt(FPR[fs])
The reciprocal of the positive square root of the value in FPR fs is approximated and placed into FPR fd. The operand
and result are values in format fmt.
The numeric accuracy of this operation is implementation dependent; it does not meet the accuracy specified by the IEEE 754 Floating Point standard. The computed result differs from both the exact result and the IEEE-mandated representation of the exact result by no more than two units in the least-significant place (ULP).
The effect of the current FCSR rounding mode on the result is implementation dependent. Restrictions:
The fields fs and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPRE- DICTABLE.
The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE.
Availability and Compatibility:
RSQRT.S and RSQRT.D: Required in all versions of MIPS64 since MIPS64 Release 1. Not available in MIPS32 Release 1. Required in MIPS32 Release 2 and all subsequent versions of MIPS32. When required, required whenever FPU is present, whether a 32-bit or 64-bit FPU, whether in 32-bit or 64-bit FP Register Mode (FIRF64=0 or 1, StatusFR=0 or 1).
Operation:
StoreFPR(fd, fmt, 1.0 / SquareRoot(valueFPR(fs, fmt)))
Exceptions:
Coprocessor Unusable, Reserved Instruction
Floating Point Exceptions:
Inexact, Division-by-zero, Unimplemented Operation, Invalid Operation, Overflow, Underflow

SB
31
26 25 21 20 16 15
655 16
Format: SB rt, offset(base) Purpose: Store Byte
To store a byte to memory.
IStore Byte 0
MIPS32
SB 101000
base
rt
offset
Description: memory[GPR[base]  offset]  GPR[rt]
The least-significant 8-bit byte of GPR rt is stored in memory at the location specified by the effective address. The
16-bit signed offset is added to the contents of GPR base to form the effective address. Restrictions:
None
Operation:
vAddr  sign_extend(offset)  GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE)
pAddr  pAddr || (pAddr xor ReverseEndian2) PSIZE-1..2 1..0
bytesel  vAddr
dataword  GPR[rt]
StoreMemory (CCA, BYTE, dataword, pAddr, vAddr, DATA)
Exceptions:
TLB Refill, TLB Invalid, TLB Modified, Bus Error, Address Error, Watch
xor BigEndianCPU2
1..0
31–8*bytesel..0
|| 08*bytesel
The MIPS32® Instruction Set Manual, Revision 6.06 344

SB Store Byte
345 The MIPS32® Instruction Set Manual, Revision 6.06

SBE
IStore Byte EVA
31
26 25 21 20 16 15 7 6 5 0
655916
Format: SBE rt, offset(base) MIPS32 Purpose: Store Byte EVA
To store a byte to user mode virtual address space when executing in kernel mode.
Description: memory[GPR[base]  offset]  GPR[rt]
The least-significant 8-bit byte of GPR rt is stored in memory at the location specified by the effective address. The
9-bit signed offset is added to the contents of GPR base to form the effective address.
The SBE instruction functions the same as the SB instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment config- ured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Enhanced Virtual Addressing section for additional information.
Implementation of this instruction is specified by the Config5EVA field being set to 1.
Restrictions:
Only usable when access to Coprocessor0 is enabled and when accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE)
pAddr  pAddr || (pAddr xor ReverseEndian2) PSIZE-1..2 1..0
SPECIAL3 011111
base
rt
offset
0
SBE 011100
bytesel  vAddr
dataword  GPR[rt]
StoreMemory (CCA, BYTE, dataword, pAddr, vAddr, DATA)
Exceptions:
TLB Refill, TLB Invalid, Bus Error, Address Error, Watch, Reserved Instruction, Coprocessor Unusable,
xor BigEndianCPU2
1..0
31-8*bytesel..0
|| 08*bytesel
The MIPS32® Instruction Set Manual, Revision 6.06 346

SC
IStore Conditional Word
pre-Release 6
31
Release 6
26 25 21 20 16 15
655 16
SC 111000
base
rt
offset
31
0
7 6 5 0
16
MIPS32
Description: if atomic_update then memory[GPR[base] + offset]  GPR[rt], GPR[rt]  1 else GPR[rt]  0
The LL and SC instructions provide primitives to implement atomic read-modify-write (RMW) operations on syn- chronizable memory locations. In Release 5, the behavior of SC is modified when Config5LLB=1.
Release 6 (with Config5ULS =1) formalizes support for uncached LL and SC sequences, whereas the pre-Release 6 LL and SC description applies to cached (coherent/non-coherent) memory types. (The description for uncached sup- port does not modify the description for cached support and is written in a self-contained manner.)
The 32-bit word in GPR rt is conditionally stored in memory at the location specified by the aligned effective address. The signed offset is added to the contents of GPR base to form an effective address.
The SC completes the RMW sequence begun by the preceding LL instruction executed on the processor. To complete the RMW sequence atomically, the following occur:
• The 32-bit word of GPR rt is stored to memory at the location specified by the aligned effective address.
• A one, indicating success, is written into GPR rt.
Otherwise, memory is not modified and a 0, indicating failure, is written into GPR rt.
If either of the following events occurs between the execution of LL and SC, the SC fails:
• A coherent store is completed by another processor or coherent I/O module into the block of synchronizable physical memory containing the word. The size and alignment of the block is implementation-dependent, but it is at least one word and at most the minimum page size.
• A coherent store is executed between an LL and SC sequence on the same processor to the block of synchroniz- able physical memory containing the word (if Config5LLB=1; else whether such a store causes the SC to fail is not predictable).
• An ERET instruction is executed. (Release 5 includes ERETNC, which will not cause the SC to fail.)
Furthermore, an SC must always compare its address against that of the LL. An SC will fail if the aligned address of the SC does not match that of the preceding LL.
A load that executes on the processor executing the LL/SC sequence to the block of synchronizable physical memory containing the word, will not cause the SC to fail (if Config5LLB=1; else such a load may cause the SC to fail).
If any of the events listed below occurs between the execution of LL and SC, the SC may fail where it could have suc-
26 25 21 20 16 15
6559
Format: SC rt, offset(base) Purpose: Store Conditional Word
To store a word to memory to complete an atomic read-modify-write
SPECIAL3 011111
base
rt
offset
0
SC 100110
The MIPS32® Instruction Set Manual, Revision 6.06 347

SC
IStore Conditional Word ceeded, i.e., success is not predictable. Portable programs should not cause any of these events.
• A load or store executed on the processor executing the LL and SC that is not to the block of synchronizable physical memory containing the word. (The load or store may cause a cache eviction between the LL and SC that results in SC failure. The load or store does not necessarily have to occur between the LL and SC.)
• Any prefetch that is executed on the processor executing the LL and SC sequence (due to a cache eviction between the LL and SC).
• A non-coherent store executed between an LL and SC sequence to the block of synchronizable physical memory containing the word.
• The instructions executed starting with the LL and ending with the SC do not lie in a 2048-byte contiguous region of virtual memory. (The region does not have to be aligned, other than the alignment required for instruc- tion words.)
CACHE operations that are local to the processor executing the LL/SC sequence will result in unpredictable behav- iour of the SC if executed between the LL and SC, that is, they may cause the SC to fail where it could have suc- ceeded. Non-local CACHE operations (address-type with coherent CCA) may cause an SC to fail on either the local processor or on the remote processor in multiprocessor or multi-threaded systems. This definition of the effects of CACHE operations is mandated if Config5LLB=1. If Config5LLB=0, then CACHE effects are implementation-depen- dent.
The following conditions must be true or the result of the SC is not predictable—the SC may fail or succeed (if Config5LLB=1, then either success or failure is mandated, else the result is UNPREDICTABLE):
• Execution of SC must have been preceded by execution of an LL instruction.
• An RMW sequence executed without intervening events that would cause the SC to fail must use the same address in the LL and SC. The address is the same if the virtual address, physical address, and cacheability & coherency attribute are identical.
Atomic RMW is provided only for synchronizable memory locations. A synchronizable memory location is one that is associated with the state and logic necessary to implement the LL/SC semantics. Whether a memory location is synchronizable depends on the processor and system configurations, and on the memory access type used for the location:
• Uniprocessor atomicity: To provide atomic RMW on a single processor, all accesses to the location must be made with memory access type of either cached noncoherent or cached coherent. All accesses must be to one or the other access type, and they may not be mixed.
• MP atomicity: To provide atomic RMW among multiple processors, all accesses to the location must be made with a memory access type of cached coherent.
• I/O System: To provide atomic RMW with a coherent I/O system, all accesses to the location must be made with a memory access type of cached coherent. If the I/O system does not use coherent memory operations, then atomic RMW cannot be provided with respect to the I/O reads and writes.
Release 6 (with Config5ULS =1) formally defines support for uncached LL and SC with the following constraints.
• Both LL and SC must be uncached, and the address must be defined as synchronizable in the system. If the address is non-synchronizable, then this may result in UNPREDICTABLE behavior. The recommended response is that the sub-system report a Bus Error to the processor.
• The use of uncached LL and SC is applicable to any address within the supported address range of the system, or any system configuration, as long as the system implements means to monitor the sequence.
• The SC that ends the sequence may fail locally, but never succeed locally within the processor. When it does not fail locally, the SC must be issued to a “monitor” which is responsible for monitoring the address. This monitor
The MIPS32® Instruction Set Manual, Revision 6.06 348

SC

• •
IStore Conditional Word makes the final determination as to whether the SC fails or not, and communicates this to the processor that initi-
ated the sequence.
It is implementation dependent as to what form the monitor takes. It is however differentiated from cached LL and SC which rely on a coherence protocol to make the determination as to whether the sequence succeeds.
Same processor uncached (but not cached) stores will cause the sequence to fail if the store address matches that of the sequence. A cached store to the same address will cause UNPREDICTABLE behavior.
Remote cached coherent stores to the same address will cause UNPREDICTABLE behavior.
Remote cached non-coherent or uncached stores may cause the sequence to fail if they address the external mon- itor and the monitor makes this determination.
As emphasized above, it is not recommended that software mix memory access types during LL and SC sequences. That is all memory accesses must be of the same type, otherwise this may result in UNPREDICTABLE behavior.
Conditions that cause UNPREDICTABLE behavior for legacy cached LL and SC sequences may also cause such behavior for uncached sequences.
A PAUSE instruction is no-op’d when it is preceded by an uncached LL.
The semantics of an uncached LL/SC atomic operation applies to any uncached CCA including UCA (UnCached Accelerated). An implementation that supports UCA must guarantee that SC does not participate in store gathering and that it ends any gathering initiated by stores preceding the SC in program order when the SC address coincides with a gathering address.
Restrictions:
The addressed location must have a memory access type of cached noncoherent or cached coherent; if it does not, the result is UNPREDICTABLE. Release 6 (with Config5ULS =1) extends support to uncached types.
The effective address must be naturally-aligned. If either of the 2 least-significant bits of the address is non-zero, an Address Error exception occurs.
Providing misaligned support for Release 6 is not a requirement for this instruction.
Availability and Compatibility
This instruction has been recoded for Release 6.
Operation:
vAddr  sign_extend(offset) + GPR[base] if vAddr1..0 ≠ 02 then
SignalException(AddressError)
endif
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE) dataword  GPR[rt]
if LLbit then
StoreMemory (CCA, WORD, dataword, pAddr, vAddr, DATA)
endif
GPR[rt]  031 || LLbit
LLbit  0 // if Config5LLB=1, SC always clears LLbit regardless of address match.
Exceptions:
TLB Refill, TLB Invalid, TLB Modified, Address Error, Watch
The MIPS32® Instruction Set Manual, Revision 6.06 349

SC
IStore Conditional Word
Programming Notes:
LL and SC are used to atomically update memory locations, as shown below.
L1:
# load counter
# increment
# try to store, checking for atomicity
# if not atomic (0), try again
# branch-delay slot
Exceptions between the LL and SC cause SC to fail, so persistent exceptions must be avoided. Some examples of these are arithmetic operations that trap, system calls, and floating point operations that trap or require software emu- lation assistance.
LL and SC function on a single processor for cached noncoherent memory so that parallel programs can be run on uniprocessor systems that do not support cached coherent memory access types.
As shown in the instruction drawing above, Release 6 implements a 9-bit offset, whereas all release levels lower than Release 6 of the MIPS architecture implement a 16-bit offset.
LL T1, (T0) ADDI T2, T1, 1 SC T2, (T0) BEQ T2, 0, L1 NOP
The MIPS32® Instruction Set Manual, Revision 6.06 350

SCE
Store Conditional Word EVA
31
26 25 21 20 16 15 7 6 5 0
655916
Format: SCE rt, offset(base) MIPS32 Purpose: Store Conditional Word EVA
To store a word to user mode virtual memory while operating in kernel mode to complete an atomic read-modify- write.
Description: if atomic_update then memory[GPR[base] + offset]  GPR[rt], GPR[rt]  1 else GPR[rt]  0
The LL and SC instructions provide primitives to implement atomic read-modify-write (RMW) operations for syn- chronizable memory locations.
Release 6 (with Config5ULS =1) formalizes support for uncached LLE and SCE sequences. (The description for uncached support does not modify the description for cached support and is written in a self-contained manner.)
The 32-bit word in GPR rt is conditionally stored in memory at the location specified by the aligned effective address. The 9-bit signed offset is added to the contents of GPR base to form an effective address.
The SCE completes the RMW sequence begun by the preceding LLE instruction executed on the processor. To com- plete the RMW sequence atomically, the following occurs:
• The 32-bit word of GPR rt is stored to memory at the location specified by the aligned effective address.
• A 1, indicating success, is written into GPR rt.
Otherwise, memory is not modified and a 0, indicating failure, is written into GPR rt.
If either of the following events occurs between the execution of LL and SC, the SC fails:
• A coherent store is completed by another processor or coherent I/O module into the block of synchronizable physical memory containing the word. The size and alignment of the block is implementation dependent, but it is at least one word and at most the minimum page size.
• An ERET instruction is executed.
If either of the following events occurs between the execution of LLE and SCE, the SCE may succeed or it may fail; the success or failure is not predictable. Portable programs should not cause one of these events.
• A memory access instruction (load, store, or prefetch) is executed on the processor executing the LLE/SCE.
• The instructions executed starting with the LLE and ending with the SCE do not lie in a 2048-byte contiguous region of virtual memory. (The region does not have to be aligned, other than the alignment required for instruc- tion words.)
The following conditions must be true or the result of the SCE is UNPREDICTABLE:
• Execution of SCE must have been preceded by execution of an LLE instruction.
• An RMW sequence executed without intervening events that would cause the SCE to fail must use the same address in the LLE and SCE. The address is the same if the virtual address, physical address, and cacheability & coherency attribute are identical.
Atomic RMW is provided only for synchronizable memory locations. A synchronizable memory location is one that is associated with the state and logic necessary to implement the LLE/SCE semantics. Whether a memory location is synchronizable depends on the processor and system configurations, and on the memory access type used for the location:
SPECIAL3 011111
base
rt
offset
0
SCE 011110
351
The MIPS32® Instruction Set Manual, Revision 6.06

SCE
IStore Conditional Word EVA

• •
Uniprocessor atomicity: To provide atomic RMW on a single processor, all accesses to the location must be made with memory access type of either cached non coherent or cached coherent. All accesses must be to one or the other access type, and they may not be mixed.
MP atomicity: To provide atomic RMW among multiple processors, all accesses to the location must be made with a memory access type of cached coherent.
I/O System: To provide atomic RMW with a coherent I/O system, all accesses to the location must be made with a memory access type of cached coherent. If the I/O system does not use coherent memory operations, then atomic RMW cannot be provided with respect to the I/O reads and writes.
The SCE instruction functions the same as the SC instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment config- ured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Enhanced Virtual Addressing section for additional information.
Implementation of this instruction is specified by the Config5EVA field being set to 1.
The definition for SCE is extended for uncached memory types in a manner identical to SC. The extension is defined
in the SC instruction description.
Restrictions:
The addressed location must have a memory access type of cached non coherent or cached coherent; if it does not, the result is UNPREDICTABLE. Release 6 (with Config5ULS =1) extends support to uncached types.
The effective address must be naturally-aligned. If either of the 2 least-significant bits of the address is non-zero, an Address Error exception occurs.
Providing misaligned support for Release 6 is not a requirement for this instruction.
Operation:
vAddr  sign_extend(offset) + GPR[base] if vAddr1..0 ≠ 02 then
SignalException(AddressError)
endif
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE) dataword  GPR[rt]
if LLbit then
StoreMemory (CCA, WORD, dataword, pAddr, vAddr, DATA)
endif
GPR[rt]  031 || LLbit Exceptions:
TLB Refill, TLB Invalid, TLB Modified, Address Error, Watch, Reserved Instruction, Coprocessor Unusable
Programming Notes:
LLE and SCE are used to atomically update memory locations, as shown below.
L1:
# load counter
# increment
# try to store, checking for atomicity
# if not atomic (0), try again
# branch-delay slot
Exceptions between the LLE and SCE cause SCE to fail, so persistent exceptions must be avoided. Examples are arithmetic operations that trap, system calls, and floating point operations that trap or require software emulation
LLE T1, (T0) ADDI T2, T1, 1 SCE T2, (T0) BEQ T2, 0, L1 NOP
The MIPS32® Instruction Set Manual, Revision 6.06 352

SCE Store Conditional Word EVA assistance.
LLE and SCE function on a single processor for cached non coherent memory so that parallel programs can be run on uniprocessor systems that do not support cached coherent memory access types.
353 The MIPS32® Instruction Set Manual, Revision 6.06

SCWP IStore Conditional Word Paired
31
26 25 21 20 16 15 11 10 7 6 5 0
6555416
Format: SCWP rt, rd, (base) MIPS32Release 6
Purpose: Store Conditional Word Paired
Conditionally store a paired word to memory to complete an atomic read-modify-write.
Description: if atomic_update then memory[GPR[base]]  {GPR[rd],GPR[rt]}, GPR[rt]  1 else GPR[rt]  0
The LLWP and SCWP instructions provide primitives to implement a paired word atomic read-modify-write (RMW) operation at a synchronizable memory location.
Release 6 (with Config5ULS =1) formalizes support for uncached LLWP and SCWP sequences. (The description for uncached support does not modify the description for cached support and is written in a self-contained manner.)
A paired word is formed from the concatenation of GPR rd and GPR rt. GPR rd is the most-significant word of the paired word, and GPR rt is the least-significant word of the paired word. Thepaired word is conditionally stored in memory at the location specified by the double-word aligned effective address from GPR base.
A paired word read or write occurs as a pair of word reads or writes that is double-word atomic. The instruction has no offset. The effective address is equal to the contents of GPR base.
The SCWP completes the RMW sequence begun by the preceding LLWP instruction executed on the processor. To complete the RMW sequence atomically, the following occur:
• The paired word formed from the concatenation of GPRs rd and rt is stored to memory at the location specified by the double-word aligned effective address.
• A one, indicating success, is written into GPR rt.
Otherwise, memory is not modified and a 0, indicating failure, is written into GPR rt.
Though legal programming requires LLWP to start the atomic read-modify-write sequence and SCWP to end the same sequence, whether the SCWP completes is only dependent on the state of LLbit and LLAddr, which are set by a preceding load-linked instruction of any type. Software must assume that pairing load-linked and store-conditional instructions in an inconsistent manner causes UNPREDICTABLE behavior.
The SCWP must always compare its double-word aligned address against that of the preceding LLWP. The SCWP will fail if the address does not match that of the preceding LLWP.
Events that occur between the execution of load-linked and store-conditional instruction types that must cause the sequence to fail are given in the legacy SC instruction description.
Additional events that occur between the execution of load-linked and store-conditional instruction types that may cause success of the sequence to be UNPREDICTABLE are defined in the SC instruction description.
A load that executes on the processor executing the LLWP/SCWP sequence to the block of synchronizable physical memory containing the paired word, will not cause the SCWP to fail.
Effect of CACHE operations, both local and remote, on a paired word atomic operation are defined in the SC instruc- tion description.
SPECIAL3 011111
base
rt
rd
0 0000
1
SC 100110
The MIPS32® Instruction Set Manual, Revision 6.06 354

355
The MIPS32® Instruction Set Manual, Revision 6.06
SCWP Store Conditional Word Paired
Atomic RMW is provided only for synchronizable memory locations. A synchronizable memory location is one that is associated with the state and logic necessary to implement the LL/SC semantics. Whether a memory location is synchronizable depends on the processor and system configurations, and on the memory access type used for the location. Requirements for Uniprocessor, MP and I/O atomicity are given in the SC definition.
The definition for SCWP is extended for uncached memory types in a manner identical to SC. The extension is defined in the SC instruction description.
Restrictions:
Load-Linked and Store-Conditional instruction types require that the addressed location must have a memory access type of cached noncoherent or cached coherent, that is the processor must have a cache. If it does not, the result is UNPREDICTABLE. Release 6 (with Config5ULS =1) extends support to uncached types.
The architecture optionally allows support for Load-Linked and Store-Conditional instruction types in a cacheless processor. Support for cacheless operation is implementation dependent. In this case, LLAddr is optional.
Providing misaligned support is not a requirement for this instruction.
Availability and Compatibility
This instruction is introduced by Release 6. It is only present if Config5XNP=0. Operation:
vAddr  GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE) datadoubleword31..0  GPR[rt]
datadoubleword63..32  GPR[rd]
if (LLbit && (pAddr == LLAddr))then
// PAIREDWORD: two word data-type that is double-word atomic
StoreMemory (CCA, PAIREDWORD, datadoubleword, pAddr, vAddr, DATA)
GPR[rt]  031 || 1’b1 else
GPR[rt]  032 endif
LLbit  0 Exceptions:
TLB Refill, TLB Invalid, TLB Modified, Reserved Instruction, Address Error, Watch
Programming Notes:
LLWP and SCWP are used to atomically update memory locations, as shown below.
L1:
U32: ADDI
ADDI
SCWP
BEQC
LLWP
BOVC
ADDI
SCWP
BEQC
T2,T3, (T0) T2, 1, U32 T2, T2, 1 T2, T3, (T0) T2, 0, L1
T2, T2, 1
T3, T3, 1
T2, T3, (T0)
T2, 0, L1
# load T2 and T3
# check whether least-significant word may overflow
# increment lower – only
# store T2 and T3
# if not atomic (0), try again
# increment lower
# increment upper
# if not atomic (0), try again

SCWP IStore Conditional Word Paired
Exceptions between the LLWP and SCWP cause SC to fail, so persistent exceptions must be avoided. Some examples of these are arithmetic operations that trap, system calls, and floating point operations that trap or require software emulation assistance.
LLWP and SCWP function on a single processor for cached noncoherent memory so that parallel programs can be run on uniprocessor systems that do not support cached coherent memory access types.
The MIPS32® Instruction Set Manual, Revision 6.06 356

SCWP Store Conditional Word Paired
357 The MIPS32® Instruction Set Manual, Revision 6.06

SCWPE IStore Conditional Word Paired EVA
31
26 25 21 20 16 15 11 10 7 6 5 0
6555416
Format: SCWPE rt, rd, (base) MIPS32Release 6
Purpose: Store Conditional Word Paired EVA
Conditionally store a paired word to memory to complete an atomic read-modify-write. The store occurs in kernel
mode to user virtual address space.
Description: if atomic_update then memory[GPR[base]] {GPR[rd],GPR[rt]}, GPR[rt]  1 else GPR[rt]  0
The LLWPE and SCWPE instructions provide primitives to implement a paired word atomic read-modify-write (RMW) operation at a synchronizable memory location.
Release 6 (with Config5ULS =1) formalizes support for uncached LLWPE and SCWPE sequences. (The description for uncached support does not modify the description for cached support and is written in a self-contained manner.)
A paired word is formed from the concatentation of GPR rd and GPR rt. GPR rd is the most-significant word of the double-word, and GPR rt is the least-significant word of the double-word. Thepaired word is conditionally stored in memory at the location specified by the double-word aligned effective address from GPR base.
A paired word read or write occurs as a pair of word reads or writes that is double-word atomic. The instruction has no offset. The effective address is equal to the contents of GPR base.
The SCWPE completes the RMW sequence begun by the preceding LLWPE instruction executed on the processor. To complete the RMW sequence atomically, the following occur:
• The paired word formed from the concatenation of GPRs rd and rt is stored to memory at the location specified by the double-word aligned effective address.
• A one, indicating success, is written into GPR rt.
Otherwise, memory is not modified and a 0, indicating failure, is written into GPR rt.
Though legal programming requires LLWPE to start the atomic read-modify-write sequence and SCWPE to end the same sequence, whether the SCWPE completes is only dependent on the state of LLbit and LLAddr, which are set by a preceding load-linked instruction of any type. Software must assume that pairing load-linked and store-condi- tional instructions in an inconsistent manner causes UNPREDICTABLE behavior.
The SCWPE must always compare its double-word aligned address against that of the preceding LLWPE. The SCWPE will fail if the address does not match that of the preceding LLWPE.
The SCWPE instruction functions the same as the SCWP instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment configured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also acces- sible. Refer to Volume III, Segmentation Control for additional information.
Events that occur between the execution of load-linked and store-conditional instruction types that must cause the sequence to fail are given in the legacy SC instruction definition..
Additional events that occur between the execution of load-linked and store-conditional instruction types that may cause success of the sequence to be UNPREDICTABLE are defined in the SC instruction definition.
A load that executes on the processor executing the LLWPE/SCWPE sequence to the block of synchronizable physi- cal memory containing the paired word, will not cause the SCWPE to fail.
SPECIAL3 011111
base
rt
rd
0 0000
1
SCE 011110
The MIPS32® Instruction Set Manual, Revision 6.06 358

SCWPE IStore Conditional Word Paired EVA Effect of CACHE operations, both local and remote, on a paired word atomic operation are defined in the SC instruc-
tion definition.
Atomic RMW is provided only for synchronizable memory locations. A synchronizable memory location is one that is associated with the state and logic necessary to implement the LL/SC semantics. Whether a memory location is synchronizable depends on the processor and system configurations, and on the memory access type used for the location. Requirements for Uniprocessor, MP and I/O atomicity are given in the SC definition.
The definition for SCWPE is extended for uncached memory types in a manner identical to SC. The extension is defined in the SC instruction description.
Restrictions:
Load-Linked and Store-Conditional instruction types require that the addressed location must have a memory access type of cached noncoherent or cached coherent, that is the processor must have a cache. If it does not, the result is UNPREDICTABLE. Release 6 (with Config5ULS =1) extends support to uncached types.
The architecture optionally allows support for Load-Linked and Store-Conditional instruction types in a cacheless processor. Support for cacheless operation is implementation dependent. In this case, LLAddr is optional.
Providing misaligned support is not a requirement for this instruction.
Availability and Compatibility
This instruction is introduced by Release 6. It is only present if Config5XNP=0 and Config5EVA=1. Operation:
vAddr  GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE) datadoubleword31..0  GPR[rt]
datadoubleword63..32  GPR[rd]
if (LLbit && (pAddr == LLAddr))then
// PAIREDWORD: two word data-type that is double-word atomic StoreMemory (CCA, PAIREDWORD, datadoubleword, pAddr, vAddr, DATA) GPR[rt]  031 || 1’b1
else
GPR[rt]  032
endif LLbit  0
Exceptions:
TLB Refill, TLB Invalid, TLB Modified, Reserved Instruction, Address Error, Watch, Coprocessor Unusable.
Programming Notes:
LLWPE and SCWPE are used to atomically update memory locations, as shown below.
L1:
U32:
ADDI T2, T2, 1 ADDI T3, T3, 1
LLWPE T2, T3,(T0) BOVC T2, 1, U32 ADDI T2, T2, 1 SCWPE T2, T3, (T0) BEQC T2, 0, L1
# load T2 and T3
# check whether least-significant word may overflow
# increment lower – only
# store T2 and T3
# if not atomic (0), try again
# increment lower
# increment upper
The MIPS32® Instruction Set Manual, Revision 6.06 359

SCWPE IStore Conditional Word Paired EVA SCWPE T2, T3, (T0)
BEQC T2, 0, L1 # if not atomic (0), try again
Exceptions between the LLWPE and SCWPE cause SC to fail, so persistent exceptions must be avoided. Some exam- ples of these are arithmetic operations that trap, system calls, and floating point operations that trap or require soft- ware emulation assistance.
LLWPE and SCWPE function on a single processor for cached noncoherent memory so that parallel programs can be run on uniprocessor systems that do not support cached coherent memory access types.
The MIPS32® Instruction Set Manual, Revision 6.06 360

SDBBP ISoftware Debug Breakpoint pre-Release 6
31 26 25 6 5 0
6 20 6
SPECIAL2 011100
code – use syscall
SDBBP 111111
Release 6
31
26 25 6 5 0
6 20 6
Format: SDBBP code EJTAG Purpose: Software Debug Breakpoint
To cause a debug breakpoint exception
Description:
This instruction causes a debug exception, passing control to the debug exception handler. If the processor is execut- ing in Debug Mode when the SDBBP instruction is executed, the exception is a Debug Mode Exception, which sets the DebugDExcCode field to the value 0x9 (Bp). The code field can be used for passing information to the debug exception handler, and is retrieved by the debug exception handler only by loading the contents of the memory word containing the instruction, using the DEPC register. The CODE field is not used in any way by the hardware.
Restrictions:
Availability and Compatibility:
This instruction has been recoded for Release 6.
Operation:
if Config5.SBRI=1 then /* SBRI is a MIPS Release 6 feature */
SignalException(ReservedInstruction) endif
If DebugDM = 1 then SignalDebugModeBreakpointException() endif // nested SignalDebugBreakpointException() // normal
Exceptions:
Debug Breakpoint Exception Debug Mode Breakpoint Exception
Programming Notes:
Release 6 changes the instruction encoding. The primary opcode changes from SPECIAL2 to SPECIAL. Also it defines a different function field value for SDBBP.
SPECIAL 000000
code – use syscall
SDBBP 001110
The MIPS32® Instruction Set Manual, Revision 6.06 361

SDC1
Store Doubleword from Floating Point
31
26 25 21 20 16 15 0
655 16
Format: SDC1 ft, offset(base) MIPS32 Purpose: Store Doubleword from Floating Point
To store a doubleword from an FPR to memory.
Description: memory[GPR[base] + offset]  FPR[ft]
The 64-bit doubleword in FPR ft is stored in memory at the location specified by the aligned effective address. The
16-bit signed offset is added to the contents of GPR base to form the effective address.
Restrictions:
Pre-Release 6: An Address Error exception occurs if EffectiveAddress2..0 ≠ 0 (not doubleword-aligned). Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation(vAddr, DATA, STORE) datadoubleword  ValueFPR(ft, UNINTERPRETED_DOUBLEWORD)
paddr  paddr xor ((BigEndianCPU xor ReverseEndian) || 02) StoreMemory(CCA, WORD, datadoubleword31..0, pAddr, vAddr, DATA) paddr  paddr xor 0b100
StoreMemory(CCA, WORD, datadoubleword63..32, pAddr, vAddr+4, DATA)
Exceptions:
Coprocessor Unusable, Reserved Instruction, TLB Refill, TLB Invalid, TLB Modified, Address Error, Watch
SDC1 111101
base
ft
offset
362
The MIPS32® Instruction Set Manual, Revision 6.06

SDC2 IStore Doubleword from Coprocessor 2 pre-Release 6
31 26 25 21 20 16 15 0
655 16
SDC2 111110
base
rt
offset
Release 6
31
26 25 21 20 16 15 11 10 0
6 5 5 5 11
Format: SDC2 rt, offset(base) MIPS32 Purpose: Store Doubleword from Coprocessor 2
To store a doubleword from a Coprocessor 2 register to memory
Description: memory[GPR[base] + offset]  CPR[2,rt,0]
The 64-bit doubleword in Coprocessor 2 register rt is stored in memory at the location specified by the aligned effec-
tive address. The 16-bit signed offset is added to the contents of GPR base to form the effective address.
Restrictions:
Pre-Release 6: An Address Error exception occurs if EffectiveAddress2..0 ≠ 0 (not doubleword-aligned). Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Availability and Compatibility:
This instruction has been recoded for Release 6.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation(vAddr, DATA, STORE)
lsw  CPR[2,rt,0]
msw  CPR[2,rt+1,0]
paddr  paddr xor ((BigEndianCPU xor ReverseEndian) || 02) StoreMemory(CCA, WORD, lsw, pAddr, vAddr, DATA)
paddr  paddr xor 0b100
StoreMemory(CCA, WORD, msw, pAddr, vAddr+4, DATA)
Exceptions:
Coprocessor Unusable, Reserved Instruction, TLB Refill, TLB Invalid, TLB Modified, Address Error, Watch
Programming Notes:
As shown in the instruction drawing above, Release 6 implements an 11-bit offset, whereas all release levels lower than Release 6 of the MIPS architecture implement a 16-bit offset.
COP2 010010
SDC2 01111
rt
base
offset
The MIPS32® Instruction Set Manual, Revision 6.06 363

SDXC1 Store Doubleword Indexed from Floating Point
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: SDXC1 fs, index(base) MIPS64, MIPS32 Release 2, removed in Release 6 Purpose: Store Doubleword Indexed from Floating Point
To store a doubleword from an FPR to memory (GPR+GPR addressing).
Description: memory[GPR[base]  GPR[index]]  FPR[fs]
The 64-bit doubleword in FPR fs is stored in memory at the location specified by the aligned effective address. The
contents of GPR index and GPR base are added to form the effective address. Restrictions:
An Address Error exception occurs if EffectiveAddress2..0 ≠ 0 (not doubleword-aligned).
Availability and Compatibility:
This instruction has been removed in Release 6.
Required in all versions of MIPS64 since MIPS64 Release 1. Not available in MIPS32 Release 1. Required in MIPS32 Release 2 and all subsequent versions of MIPS32. When required, these instructions are to be implemented if an FPU is present either in a 32-bit or 64-bit FPU or in a 32-bit or 64-bit FP Register Mode (FIRF64=0 or 1, StatusFR=0 or 1).
Operation:
COP1X 010011
base
index
fs
0 00000
SDXC1 001001
364
The MIPS32® Instruction Set Manual, Revision 6.06
vAddr  GPR[base]  GPR[index] if vAddr  03 then
2..0 SignalException(AddressError)
endif
(pAddr, CCA)  AddressTranslation(vAddr, DATA, STORE) datadoubleword  ValueFPR(fs, UNINTERPRETED_DOUBLEWORD)
paddr  paddr xor ((BigEndianCPU xor ReverseEndian) || 02) StoreMemory(CCA, WORD, datadoubleword31..0, pAddr, vAddr, DATA) paddr  paddr xor 0b100
StoreMemory(CCA, WORD, datadoubleword63..32, pAddr, vAddr+4, DATA)
Exceptions:
TLB Refill, TLB Invalid, TLB Modified, Coprocessor Unusable, Address Error, Reserved Instruction, Watch.

SEB
ISign-Extend Byte
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: SEB rd, rt MIPS32 Release 2 Purpose: Sign-Extend Byte
To sign-extend the least significant byte of GPR rt and store the value into GPR rd. Description: GPR[rd]  SignExtend(GPR[rt]7..0)
The least significant byte from GPR rt is sign-extended and stored in GPR rd. Restrictions:
Prior to architecture Release 2, this instruction resulted in a Reserved Instruction exception.
sOperation:
sGPR[rd]  sign_extend(GPR[rt]7..0)
Exceptions:
Reserved Instruction
Programming Notes:
For symmetry with the SEB and SEH instructions, you expect that there would be ZEB and ZEH instructions that zero-extend the source operand and expect that the SEW and ZEW instructions would exist to sign- or zero-extend a word to a doubleword. These instructions do not exist because there are functionally-equivalent instructions already in the instruction set. The following table shows the instructions providing the equivalent functions.
SPECIAL3 011111
0 00000
rt
rd
SEB 10000
BSHFL 100000
Expected Instruction
ZEB rx,ry
ZEH rx,ry
Function
Zero-Extend Byte Zero-Extend Halfword
Equivalent Instruction
ANDI rx,ry,0xFF
ANDI rx,ry,0xFFFF
The MIPS32® Instruction Set Manual, Revision 6.06
365

SEH
Sign-Extend Halfword
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: SEH rd, rt MIPS32 Release 2 Purpose: Sign-Extend Halfword
To sign-extend the least significant halfword of GPR rt and store the value into GPR rd. Description: GPR[rd]  SignExtend(GPR[rt]15..0)
The least significant halfword from GPR rt is sign-extended and stored in GPR rd. Restrictions:
In implementations prior to Release 2 of the architecture, this instruction resulted in a Reserved Instruction exception.
sOperation:
sGPR[rd]  signextend(GPR[rt]15..0)
Exceptions:
Reserved Instruction
Programming Notes:
The SEH instruction can be used to convert two contiguous halfwords to sign-extended word values in three instruc- tions. For example:
SPECIAL3 011111
0 00000
rt
rd
SEH 11000
BSHFL 100000
lw t0, 0(a1) seh t1, t0 sra t0, t0, 16
/* Read two contiguous halfwords */
/* t1 = lower halfword sign-extended to word */
/* t0 = upper halfword sign-extended to word */
Zero-extended halfwords can be created by changing the SEH and SRA instructions to ANDI and SRL instructions, respectively.
For symmetry with the SEB and SEH instructions, you expect that there would be ZEB and ZEH instructions that zero-extend the source operand and expect that the SEW and ZEW instructions would exist to sign- or zero-extend a word to a doubleword. These instructions do not exist because there are functionally-equivalent instructions already in the instruction set. The following table shows the instructions providing the equivalent functions.
Expected Instruction
ZEB rx,ry
ZEH rx,ry
Function
Zero-Extend Byte Zero-Extend Halfword
Equivalent Instruction
ANDI rx,ry,0xFF
ANDI rx,ry,0xFFFF
366
The MIPS32® Instruction Set Manual, Revision 6.06

SEH ISign-Extend Halfword
The MIPS32® Instruction Set Manual, Revision 6.06 367

SEL.fmt Select floating point values with FPR condition
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: SEL.fmt
SEL.S fd,fs,ft MIPS32 Release 6
SEL.D fd,fs,ft MIPS32 Release 6 Purpose: Select floating point values with FPR condition
Description: FPR[fd]  FPR[fd].bit0 ? FPR[ft] : FPR[fs]
SEL fmt is a select operation, with a condition input in FPR fd, and 2 data inputs in FPRs ft and fs.
• If the condition is true, the value of ft is written to fd.
• If the condition is false, the value of fs is written to fd.
The condition input is specified by FPR fd, and is overwritten by the result.
The condition is true only if bit 0 of the condition input FPR fd is set. Other bits are ignored.
This instruction has floating point formats S and D, but these specify only the width of the operands. SEL.S can be used for 32-bit W data, and SEL.D can be used for 64 bit L data.
This instruction does not cause data-dependent exceptions. It does not trap on NaNs, and the FCSRCause and FCSRFlags fields are not modified.
Restrictions:
None
Availability and Compatibility:
SEL fmt is introduced by and required as of MIPS32 Release 6.
Special Considerations:
Only formats S and D are valid. Other format values may be used to encode other instructions. Unused format encod- ings are required to signal the Reserved Instruction exception.
Operation:
tmp  ValueFPR(fd, UNINTERPRETED_WORD) cond  tmp.bit0
if cond then
tmp  ValueFPR(ft, fmt) else
tmp  ValueFPR(fs, fmt) endif
StoreFPR(fd, fmt, tmp)
Exceptions:
Coprocessor Unusable, Reserved Instruction
Floating Point Exceptions:
None
COP1 010001
fmt
S, D only
ft
fs
fd
SEL 010000
368
The MIPS32® Instruction Set Manual, Revision 6.06

SELEQZ SELNEZ ISelect integer GPR value or zero
31
26 25 21 20 16 15 11 10 6 5 0
655556
SPECIAL 000000
rs
rt
rd
00000
SELEQZ 110101
SPECIAL 000000
rs
rt
rd
00000
SELNEZ 110111
Format: SELEQZ SELNEZ SELEQZ rd,rs,rt SELNEZ rd,rs,rt
Purpose: Select integer GPR value or zero Description:
SELEQZ: GPR[rd]  GPR[rt] ? 0 : GPR[rs] SELNEZ: GPR[rd]  GPR[rt] ? GPR[rs] : 0
MIPS32 Release 6 MIPS32 Release 6
• SELEQZ is a select operation, with a condition input in GPR rt, one explicit data input in GPR rs, and implicit data input 0. The condition is true only if all bits in GPR rt are zero.
• SELNEZ is a select operation, with a condition input in GPR rt, one explicit data input in GPR rs, and implicit data input 0. The condition is true only if any bit in GPR rt is nonzero
If the condition is true, the value of rs is written to rd. If the condition is false, the zero written to rd.
This instruction operates on all GPRLEN bits of the CPU registers, that is, all 32 bits on a 32-bit CPU, and all 64 bits on a 64-bit CPU. All GPRLEN bits of rt are tested.
Restrictions:
None
Availability and Compatibility:
These instructions are introduced by and required as of MIPS32 Release 6.
Special Considerations:
None
Operation:
SELNEZ: cond  GPR[rt]  0 SELEQZ: cond  GPR[rt] = 0 if cond then
tmp GPR[rs] else
tmp 0 endif
GPR[rd] tmp Exceptions:
None
The MIPS32® Instruction Set Manual, Revision 6.06 369

370
The MIPS32® Instruction Set Manual, Revision 6.06
SELEQZ SELNEZ
Select integer GPR value or zero
Programming Note:
Release 6 removes the Pre-Release 6 instructions MOVZ and MOVN:
MOVZ: if GPR[rt] = 0 then GPR[rd]  GPR[rs] MOVN: if GPR[rt] ≠ 0 then GPR[rd]  GPR[rs]
MOVZ can be emulated using Release 6 instructions as follows:
SELEQZ at, rs, rt
SELNEZ rd, rd, rt
OR rd, rd, at
Similarly MOVN:
SELNEZ at, rs, rt
SELEQZ rd, rd, rt
OR rd, rd, at
The more general select operation requires 4 registers (1 output + 3 inputs (1 condition + 2 data)) and can be expressed:
rD  if rC then rA else rB
The more general select can be created using Release 6 instructions as follows:
SELNEZ at, rB, rC
SELNEZ rD, rA, rC
OR rD, rD, at

SELEQZ.fmt SELNEQZ.fmt ISelect floating point value or zero with FPR condition.
31
26 25 21 20 16 15 11 10 6 5 0
655556
COP1 010001
fmt
S, D only
ft
fs
fd
SELEQZ 010100
COP1 010001
fmt
S, D only
ft
fs
fd
SELNEZ 010111
Format: SELEQZ.fmt SELNEQZ.fmt SELEQZ.S fd,fs,ft
SELEQZ.D fd,fs,ft
SELNEZ.S fd,fs,ft
SELNEZ.D fd,fs,ft
Purpose: Select floating point value or zero with FPR condition. Description:
SELEQZ.fmt: FPR[fd]  FPR[ft].bit0 ? 0 : FPR[fs] SELNEZ.fmt: FPR[fd]  FPR[ft].bit0 ? FPR[fs]: 0
MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6 MIPS32 Release 6
• SELEQZ.fmt is a select operation, with a condition input in FPR ft, one explicit data input in FPR fs, and implicit data input 0. The condition is true only if bit 0 of FPR ft is zero.
• SELNEZ.fmt is a select operation, with a condition input in FPR ft, one explicit data input in FPR fs, and implicit data input 0. The condition is true only if bit 0 of FPR ft is nonzero.
If the condition is true, the value of fs is written to fd.
If the condition is false, the value that has all bits zero is written to fd.
This instruction has floating point formats S and D, but these specify only the width of the operands. Format S can be used for 32-bit W data, and format D can be used for 64 bit L data. The condition test is restricted to bit 0 of FPR ft. Other bits are ignored.
This instruction has no execution exception behavior. It does not trap on NaNs, and the FCSRCause and FCSRFlags fields are not modified.
Restrictions:
FPR fd destination register bits beyond the format width are UNPREDICTABLE. For example, if fmt is S, then fd bits 0-31 are defined, but bits 32 and above are UNPREDICTABLE. If fmt is D, then fd bits 0-63 are defined.
Availability and Compatibility:
These instructions are introduced by and required as of MIPS32 Release 6.
Special Considerations:
Only formats S and D are valid. Other format values may be used to encode other instructions. Unused format encod- ings are required to signal the Reserved Instruction exception.
Operation:
tmp  ValueFPR(ft, UNINTERPRETED_WORD) SELEQZ: cond  tmp.bit0 = 0
SELNEZ: cond  tmp.bit0  0
if cond then
tmp  ValueFPR(fs, fmt) else
The MIPS32® Instruction Set Manual, Revision 6.06 371

372
The MIPS32® Instruction Set Manual, Revision 6.06
SELEQZ.fmt SELNEQZ.fmt Select floating point value or zero with FPR condition.
tmp 0 /* all bits set to zero */ endif
StoreFPR(fd, fmt, tmp)
Exceptions:
Coprocessor Unusable, Reserved Instruction
Floating Point Exceptions:

SH
31
26 25 21 20 16 15
655 16
Format: SH rt, offset(base) Purpose: Store Halfword
To store a halfword to memory.
IStore Halfword 0
MIPS32
SH 101001
base
rt
offset
Description: memory[GPR[base] + offset]  GPR[rt]
The least-significant 16-bit halfword of register rt is stored in memory at the location specified by the aligned effec-
tive address. The 16-bit signed offset is added to the contents of GPR base to form the effective address. Restrictions:
Pre-Release 6: The effective address must be naturally-aligned. If the least-significant bit of the address is non-zero, an Address Error exception occurs.
Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE)
pAddr  pAddrPSIZE-1..2 || (pAddr1..0 xor (ReverseEndian || 0))
bytesel  vAddr1..0 xor (BigEndianCPU || 0)
dataword  GPR[rt] || 08*bytesel 31-8*bytesel..0
StoreMemory (CCA, HALFWORD, dataword, pAddr, vAddr, DATA)
Exceptions:
TLB Refill, TLB Invalid, TLB Modified, Address Error, Watch
The MIPS32® Instruction Set Manual, Revision 6.06 373

SHE
Store Halfword EVA
31
26 25 21 20 16 15 7 6 5 0
655916
Format: SHE rt, offset(base) MIPS32 Purpose: Store Halfword EVA
To store a halfword to user mode virtual address space when executing in kernel mode.
Description: memory[GPR[base] + offset]  GPR[rt]
The least-significant 16-bit halfword of register rt is stored in memory at the location specified by the aligned effec-
tive address. The 9-bit signed offset is added to the contents of GPR base to form the effective address.
The SHE instruction functions the same as the SH instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment config- ured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Enhanced Virtual Addressing section for additional information.
Implementation of this instruction is specified by the Config5EVA field being set to 1.
Restrictions:
Only usable in kernel mode when accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode.
Pre-Release 6: The effective address must be naturally-aligned. If the least-significant bit of the address is non-zero, an Address Error exception occurs.
Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE)
pAddr  pAddrPSIZE-1..2 || (pAddr1..0 xor (ReverseEndian || 0))
bytesel  vAddr1..0 xor (BigEndianCPU || 0)
dataword  GPR[rt] || 08*bytesel 31-8*bytesel..0
StoreMemory (CCA, HALFWORD, dataword, pAddr, vAddr, DATA)
Exceptions:
TLB Refill, TLB Invalid, Bus Error, Address Error, Watch, Reserved Instruction, Coprocessor Unusable
SPECIAL3 011111
base
rt
offset
0
SHE 011101
374
The MIPS32® Instruction Set Manual, Revision 6.06

SIGRIE ISignal Reserved Instruction Exception
31
26 25 21 20 16 15
655 16
Format: SIGRIE code
Purpose: Signal Reserved Instruction Exception
The SIGRIE instruction signals a Reserved Instruction exception.
0
MIPS32 Release 6
REGIMM 000001
00000
SIGRIE 10111
code
Description: SignalException(ReservedInstruction)
The SIGRIE instruction signals a Reserved Instruction exception. Implementations should use exactly the same
mechanisms as they use for reserved instructions that are not defined by the Architecture. The 16-bit code field is available for software use.
Restrictions:
The 16-bit code field is available for software use. The value zero is considered the default value. Software may pro- vide extended functionality by interpreting nonzero values of the code field in a manner that is outside the scope of this architecture specification.
Availability and Compatibility:
This instruction is introduced by and required as of Release 6.
Pre-Release 6: this instruction encoding was reserved, and required to signal a Reserved Instruction exception. There- fore this instruction can be considered to be both backwards and forwards compatible.
Operation:
SignalException(ReservedInstruction)
Exceptions:
Reserved Instruction
The MIPS32® Instruction Set Manual, Revision 6.06 375

376
The MIPS32® Instruction Set Manual, Revision 6.06
SLL
Shift Word Left Logical
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: SLL rd, rt, sa MIPS32 Purpose: Shift Word Left Logical
To left-shift a word by a fixed number of bits.
Description: GPR[rd]  GPR[rt] << sa The contents of the low-order 32-bit word of GPR rt are shifted left, inserting zeros into the emptied bits. The word result is placed in GPR rd. The bit-shift amount is specified by sa. Restrictions: SPECIAL 000000 0 00000 rt rd sa SLL 000000 None Operation: ssa temp  GPR[rt] || 0s (31-s)..0 GPR[rd]  temp Exceptions: None Programming Notes: SLL r0, r0, 0, expressed as NOP, is the assembly idiom used to denote no operation. SLL r0, r0, 1, expressed as SSNOP, is the assembly idiom used to denote no operation that causes an issue break on superscalar processors. SLLV IShift Word Left Logical Variable 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: SLLV rd, rt, rs MIPS32 Purpose: Shift Word Left Logical Variable To left-shift a word by a variable number of bits. Description: GPR[rd]  GPR[rt] << GPR[rs] The contents of the low-order 32-bit word of GPR rt are shifted left, inserting zeros into the emptied bits. The result- ing word is placed in GPR rd. The bit-shift amount is specified by the low-order 5 bits of GPR rs. Restrictions: SPECIAL 000000 rs rt rd 0 00000 SLLV 000100 None Operation: s  GPR[rs]4..0 temp  GPR[rt] || 0s (31-s)..0 GPR[rd]  temp Exceptions: None Programming Notes: None The MIPS32® Instruction Set Manual, Revision 6.06 377 SLT Set on Less Than 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: SLT rd, rs, rt MIPS32 Purpose: Set on Less Than To record the result of a less-than comparison. Description: GPR[rd]  (GPR[rs] < GPR[rt]) Compare the contents of GPR rs and GPR rt as signed integers; record the Boolean result of the comparison in GPR rd. If GPR rs is less than GPR rt, the result is 1 (true); otherwise, it is 0 (false). The arithmetic comparison does not cause an Integer Overflow exception. Restrictions: None Operation: if GPR[rs] < GPR[rt] then GPR[rd]  0GPRLEN-1 || 1 else GPR[rd]  0GPRLEN endif Exceptions: None SPECIAL 000000 rs rt rd 0 00000 SLT 101010 378 The MIPS32® Instruction Set Manual, Revision 6.06 SLTI ISet on Less Than Immediate 31 26 25 21 20 16 15 0 655 16 Format: SLTI rt, rs, immediate MIPS32 Purpose: Set on Less Than Immediate To record the result of a less-than comparison with a constant. Description: GPR[rt]  (GPR[rs] < sign_extend(immediate) ) Compare the contents of GPR rs and the 16-bit signed immediate as signed integers; record the Boolean result of the comparison in GPR rt. If GPR rs is less than immediate, the result is 1 (true); otherwise, it is 0 (false). The arithmetic comparison does not cause an Integer Overflow exception. Restrictions: None Operation: if GPR[rs] < sign_extend(immediate) then GPR[rt]  0GPRLEN-1|| 1 else GPR[rt]  0GPRLEN endif Exceptions: None SLTI 001010 rs rt immediate The MIPS32® Instruction Set Manual, Revision 6.06 379 SLTIU Set on Less Than Immediate Unsigned 31 26 25 21 20 16 15 0 655 16 Format: SLTIU rt, rs, immediate MIPS32 Purpose: Set on Less Than Immediate Unsigned To record the result of an unsigned less-than comparison with a constant. Description: GPR[rt]  (GPR[rs] < sign_extend(immediate)) Compare the contents of GPR rs and the sign-extended 16-bit immediate as unsigned integers; record the Boolean result of the comparison in GPR rt. If GPR rs is less than immediate, the result is 1 (true); otherwise, it is 0 (false). Because the 16-bit immediate is sign-extended before comparison, the instruction can represent the smallest or largest unsigned numbers. The representable values are at the minimum [0, 32767] or maximum [max_unsigned-32767, max_unsigned] end of the unsigned range. The arithmetic comparison does not cause an Integer Overflow exception. Restrictions: None Operation: if (0 || GPR[rs]) < (0 || sign_extend(immediate)) then GPR[rt]  0GPRLEN-1 || 1 else GPR[rt]  0GPRLEN endif Exceptions: None SLTIU 001011 rs rt immediate 380 The MIPS32® Instruction Set Manual, Revision 6.06 SLTU ISet on Less Than Unsigned 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: SLTU rd, rs, rt MIPS32 Purpose: Set on Less Than Unsigned To record the result of an unsigned less-than comparison. Description: GPR[rd]  (GPR[rs] < GPR[rt]) Compare the contents of GPR rs and GPR rt as unsigned integers; record the Boolean result of the comparison in GPR rd. If GPR rs is less than GPR rt, the result is 1 (true); otherwise, it is 0 (false). The arithmetic comparison does not cause an Integer Overflow exception. Restrictions: None Operation: if (0 || GPR[rs]) < (0 || GPR[rt]) then GPR[rd]  0GPRLEN-1 || 1 else GPR[rd]  0GPRLEN endif Exceptions: None SPECIAL 000000 rs rt rd 0 00000 SLTU 101011 The MIPS32® Instruction Set Manual, Revision 6.06 381 SQRT.fmt Floating Point Square Root 31 26 25 21 20 16 15 11 10 6 5 655556 Format: SQRT.fmt SQRT.S fd, fs SQRT.D fd, fs Purpose: Floating Point Square Root To compute the square root of an FP value. 0 MIPS32 MIPS32 COP1 010001 fmt 0 00000 fs fd SQRT 000100 382 The MIPS32® Instruction Set Manual, Revision 6.06 Description: FPR[fd]  SQRT(FPR[fs]) The square root of the value in FPR fs is calculated to infinite precision, rounded according to the current rounding mode in FCSR, and placed into FPR fd. The operand and result are values in format fmt. If the value in FPR fs corresponds to – 0, the result is – 0. Restrictions: If the value in FPR fs is less than 0, an Invalid Operation condition is raised. The fields fs and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPRE- DICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. Operation: StoreFPR(fd, fmt, SquareRoot(ValueFPR(fs, fmt))) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Invalid Operation, Inexact, Unimplemented Operation SRA IShift Word Right Arithmetic 31 26 25 21 20 16 15 11 10 6 5 0 655556 Format: SRA rd, rt, sa MIPS32 Purpose: Shift Word Right Arithmetic To execute an arithmetic right-shift of a word by a fixed number of bits. Description: GPR[rd]  GPR[rt] >> sa (arithmetic)
The contents of the low-order 32-bit word of GPR rt are shifted right, duplicating the sign-bit (bit 31) in the emptied
bits; the word result is placed in GPR rd. The bit-shift amount is specified by sa. Restrictions:
SPECIAL 000000
0 00000
rt
rd
sa
SRA 000011
None sOperation:
sssa
temp  GPR[rt] )s || GPR[rt]
31..s
31 GPR[rd]  temp
Exceptions:
None
The MIPS32® Instruction Set Manual, Revision 6.06
383

SRAV Shift Word Right Arithmetic Variable
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: SRAV rd, rt, rs MIPS32 Purpose: Shift Word Right Arithmetic Variable
To execute an arithmetic right-shift of a word by a variable number of bits.
Description: GPR[rd]  GPR[rt] >> GPR[rs] (arithmetic)
The contents of the low-order 32-bit word of GPR rt are shifted right, duplicating the sign-bit (bit 31) in the emptied
bits; the word result is placed in GPR rd. The bit-shift amount is specified by the low-order 5 bits of GPR rs. Restrictions:
SPECIAL 000000
rs
rt
rd
0 00000
SRAV 000111
384
The MIPS32® Instruction Set Manual, Revision 6.06
None Operation:
s  GPR[rs]4..0
temp  (GPR[rt] )s || GPR[rt]
GPR[rd]  temp
Exceptions:
None
31..s
31

SRL
IShift Word Right Logical
31
26 25 22 21 20 16 15 11 10 6 5 0
6415556
Format: SRL rd, rt, sa MIPS32 Purpose: Shift Word Right Logical
To execute a logical right-shift of a word by a fixed number of bits.
Description: GPR[rd]  GPR[rt] >> sa (logical)
The contents of the low-order 32-bit word of GPR rt are shifted right, inserting zeros into the emptied bits. The word
result is placed in GPR rd. The bit-shift amount is specified by sa. Restrictions:
None sOperation:
SPECIAL 000000
0000
R 0
rt
rd
sa
SRL 000010
sssa
temp  0s || GPR[rt] GPR[rd]  temp
Exceptions:
None
31..s
The MIPS32® Instruction Set Manual, Revision 6.06 385

SRLV
Shift Word Right Logical Variable
31
26 25 21 20 16 15 11 10 7 6 5 0
SPECIAL 000000
rs
rt
rd
0000
R 0
SRLV 000110
386
The MIPS32® Instruction Set Manual, Revision 6.06
6555416
Format: SRLV rd, rt, rs
Purpose: Shift Word Right Logical Variable
To execute a logical right-shift of a word by a variable number of bits.
MIPS32
Description: GPR[rd]  GPR[rt] >> GPR[rs] (logical)
The contents of the low-order 32-bit word of GPR rt are shifted right, inserting zeros into the emptied bits; the word
result is placed in GPR rd. The bit-shift amount is specified by the low-order 5 bits of GPR rs. Restrictions:
None Operation:
s  GPR[rs]4..0
temp  0s || GPR[rt] GPR[rd]  temp
Exceptions:
None
31..s

SSNOP ISuperscalar No Operation
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: SSNOP AssemblyIdiomMIPS32 Purpose: Superscalar No Operation
Break superscalar issue on a superscalar processor.
Description:
SSNOP is the assembly idiom used to denote superscalar no operation. The actual instruction is interpreted by the hardware as SLL r0, r0, 1.
This instruction alters the instruction issue behavior on a superscalar processor by forcing the SSNOP instruction to single-issue. The processor must then end the current instruction issue between the instruction previous to the SSNOP and the SSNOP. The SSNOP then issues alone in the next issue slot.
On a single-issue processor, this instruction is a NOP that takes an issue slot.
Restrictions:
None
Availability and Compatibility
Release 6: the special no-operation instruction SSNOP is deprecated: it behaves the same as a conventional NOP. Its special behavior with respect to instruction issue is no longer guaranteed. The EHB and JR.HB instructions are pro- vided to clear execution and instruction hazards.
Assemblers targeting specifically Release 6 should reject the SSNOP instruction with an error.
Operation:
None
Exceptions:
None
Programming Notes:
SSNOP is intended for use primarily to allow the programmer control over CP0 hazards by converting instructions into cycles in a superscalar processor. For example, to insert at least two cycles between an MTC0 and an ERET, one would use the following sequence:
mtc0 x,y ssnop ssnop eret
The MTC0 issues in cycle T. Because the SSNOP instructions must issue alone, they may issue no earlier than cycle T+1 and cycle T+2, respectively. Finally, the ERET issues no earlier than cycle T+3. Although the instruction after an SSNOP may issue no earlier than the cycle after the SSNOP is issued, that instruction may issue later. This is because other implementation-dependent issue rules may apply that prevent an issue in the next cycle. Processors should not introduce any unnecessary delay in issuing SSNOP instructions.
SPECIAL 000000
0 00000
0 00000
0 00000
1 00001
SLL 000000
The MIPS32® Instruction Set Manual, Revision 6.06 387

SUB
Subtract Word
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: SUB rd, rs, rt MIPS32 Purpose: Subtract Word
To subtract 32-bit integers. If overflow occurs, then trap.
Description: GPR[rd]  GPR[rs]  GPR[rt]
The 32-bit word value in GPR rt is subtracted from the 32-bit value in GPR rs to produce a 32-bit result. If the sub- traction results in 32-bit 2’s complement arithmetic overflow, then the destination register is not modified and an Inte- ger Overflow exception occurs. If it does not overflow, the 32-bit result is placed into GPR rd.
Restrictions:
None
Operation:
temp  (GPR[rs]31||GPR[rs]31..0)  (GPR[rt]31||GPR[rt]31..0) if temp32 ≠ temp31 then
SignalException(IntegerOverflow)
else
GPR[rd]  temp31..0 endif
Exceptions:
Integer Overflow
Programming Notes:
SUBU performs the same arithmetic operation but does not trap on overflow.
SPECIAL 000000
rs
rt
rd
0 00000
SUB 100010
388
The MIPS32® Instruction Set Manual, Revision 6.06

SUB.fmt IFloating Point Subtract
31
26 25 21 20 16 15 11 10 6 5
655556
COP1 010001
fmt
ft
fs
fd
SUB 000001
Format: SUB.fmt
SUB.S fd, fs, ft
SUB.D fd, fs, ft
SUB.PS fd, fs, ft
Purpose: Floating Point Subtract To subtract FP values.
Description: FPR[fd]  FPR[fs]  FPR[ft]
0
MIPS32
MIPS32 MIPS64,MIPS32 Release 2, removed in Release 6
The value in FPR ft is subtracted from the value in FPR fs. The result is calculated to infinite precision, rounded according to the current rounding mode in FCSR, and placed into FPR fd. The operands and result are values in for- mat fmt. SUB.PS subtracts the upper and lower halves of FPR fs and FPR ft independently, and ORs together any gen- erated exceptional conditions.
Restrictions:
The fields fs, ft, and fd must specify FPRs valid for operands of type fmt. If the fields are not valid, the result is UNPREDICTABLE.
The operands must be values in format fmt; if they are not, the result is UNPREDICTABLE and the value of the operand FPRs becomes UNPREDICTABLE.
The result of SUB.PS is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model; it is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU.
Availability and Compatibility:
SUB.PS has been removed in Release 6.
Operation:
StoreFPR (fd, fmt, ValueFPR(fs, fmt) fmt ValueFPR(ft, fmt)) CPU Exceptions:
Coprocessor Unusable, Reserved Instruction
FPU Exceptions:
Inexact, Overflow, Underflow, Invalid Op, Unimplemented Op
The MIPS32® Instruction Set Manual, Revision 6.06 389

SUBU Subtract Unsigned Word
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: SUBU rd, rs, rt MIPS32 Purpose: Subtract Unsigned Word
To subtract 32-bit integers.
Description: GPR[rd]  GPR[rs]  GPR[rt]
The 32-bit word value in GPR rt is subtracted from the 32-bit value in GPR rs and the 32-bit arithmetic result is and
placed into GPR rd.
No integer overflow exception occurs under any circumstances.
Restrictions:
None
Operation:
temp  GPR[rs]  GPR[rt] GPR[rd]  temp
Exceptions:
None
Programming Notes:
The term “unsigned” in the instruction name is a misnomer; this operation is 32-bit modulo arithmetic that does not trap on overflow. It is appropriate for unsigned arithmetic, such as address arithmetic, or integer arithmetic environ- ments that ignore overflow, such as C language arithmetic.
SPECIAL 000000
rs
rt
rd
0 00000
SUBU 100011
390
The MIPS32® Instruction Set Manual, Revision 6.06

SUXC1 IStore Doubleword Indexed Unaligned from Floating Point
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: SUXC1 fs, index(base) MIPS64,MIPS32 Release 2, removed in Release 6 Purpose: Store Doubleword Indexed Unaligned from Floating Point
To store a doubleword from an FPR to memory (GPR+GPR addressing) ignoring alignment.
Description: memory[(GPR[base] + GPR[index])PSIZE-1..3]  FPR[fs]
The contents of the 64-bit doubleword in FPR fs is stored at the memory location specified by the effective address. The contents of GPR index and GPR base are added to form the effective address. The effective address is double- word-aligned; EffectiveAddress2..0 are ignored.
Restrictions:
The result of this instruction is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model. The instruction is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU.
Availability and Compatibility
This instruction has been removed in Release 6.
Operation:
COP1X 010011
base
index
fs
0 00000
SUXC1 001101
vAddr  (GPR[base]+GPR[index])
(pAddr, CCA)  AddressTranslation(vAddr, DATA, STORE) datadoubleword  ValueFPR(fs, UNINTERPRETED_DOUBLEWORD)
paddr  paddr xor ((BigEndianCPU xor ReverseEndian) || 02) StoreMemory(CCA, WORD, datadoubleword31..0, pAddr, vAddr, DATA) paddr  paddr xor 0b100
StoreMemory(CCA, WORD, datadoubleword63..32, pAddr, vAddr+4, DATA)
Exceptions:
Coprocessor Unusable, Reserved Instruction, TLB Refill, TLB Invalid, TLB Modified, Watch
63..3
|| 03
The MIPS32® Instruction Set Manual, Revision 6.06 391

SW
Store Word
31
26 25 21 20 16 15
655 16
Format: SW rt, offset(base) Purpose: Store Word
To store a word to memory.
0
MIPS32
SW 101011
base
rt
offset
392
The MIPS32® Instruction Set Manual, Revision 6.06
Description: memory[GPR[base] + offset]  GPR[rt]
The least-significant 32-bit word of GPR rt is stored in memory at the location specified by the aligned effective
address. The 16-bit signed offset is added to the contents of GPR base to form the effective address. Restrictions:
Pre-Release 6: The effective address must be naturally-aligned. If either of the 2 least-significant bits of the address is non-zero, an Address Error exception occurs.
Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE) dataword  GPR[rt]
StoreMemory (CCA, WORD, dataword, pAddr, vAddr, DATA)
Exceptions:
TLB Refill, TLB Invalid, TLB Modified, Address Error, Watch

SWC1 IStore Word from Floating Point
31
26 25 21 20 16 15 0
655 16
SWC1 ft, offset(base) MIPS32 Purpose: Store Word from Floating Point
To store a word from an FPR to memory.
Description: memory[GPR[base] + offset]  FPR[ft]
The low 32-bit word from FPR ft is stored in memory at the location specified by the aligned effective address. The
16-bit signed offset is added to the contents of GPR base to form the effective address. Restrictions:
Pre-Release 6: An Address Error exception occurs if EffectiveAddress1..0 ≠ 0 (not word-aligned).
Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation(vAddr, DATA, STORE) dataword  ValueFPR(ft, UNINTERPRETED_WORD) StoreMemory(CCA, WORD, dataword, pAddr, vAddr, DATA)
Exceptions:
Coprocessor Unusable, Reserved Instruction, TLB Refill, TLB Invalid, TLB Modified, Address Error, Watch
SWC1 111001
base
ft
offset
The MIPS32® Instruction Set Manual, Revision 6.06 393

SWC2 Store Word from Coprocessor 2 pre-Release 6
31 26 25 21 20 16 15 0
655 16
SWC2 111010
base
rt
offset
Release 6
31
26 25 21 20 16 15 11 10 0
6 5 5 5 11
Format: SWC2 rt, offset(base) MIPS32 Purpose: Store Word from Coprocessor 2
To store a word from a COP2 register to memory
Description: memory[GPR[base] + offset]  CPR[2,rt,0]
The low 32-bit word from COP2 (Coprocessor 2) register rt is stored in memory at the location specified by the
aligned effective address. The signed offset is added to the contents of GPR base to form the effective address. Restrictions:
Pre-Release 6: An Address Error exception occurs if EffectiveAddress1..0 ≠ 0 (not word-aligned).
Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Availability and Compatibility
This instruction has been recoded for Release 6.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation(vAddr, DATA, STORE) dataword  CPR[2,rt,0]
StoreMemory(CCA, WORD, dataword, pAddr, vAddr, DATA)
Exceptions:
Coprocessor Unusable, Reserved Instruction, TLB Refill, TLB Invalid, TLB Modified, Address Error, Watch
Programming Notes:
As shown in the instruction drawing above, Release 6 implements an 11-bit offset, whereas all release levels lower than Release 6 of the MIPS architecture implement a 16-bit offset.
COP2 010010
SWC2 01011
rt
base
offset
394
The MIPS32® Instruction Set Manual, Revision 6.06

SWC2 IStore Word from Coprocessor 2
The MIPS32® Instruction Set Manual, Revision 6.06 395

SWE
Store Word EVA
31
26 25 21 20 16 15 7 6 5 0
SPECIAL3 011111
base
rt
offset
0
SWE 011111
396
The MIPS32® Instruction Set Manual, Revision 6.06
655916
Format: SWE rt, offset(base) Purpose: Store Word EVA
To store a word to user mode virtual address space when executing in kernel mode.
MIPS32
Description: memory[GPR[base] + offset]  GPR[rt]
The least-significant 32-bit word of GPR rt is stored in memory at the location specified by the aligned effective
address. The 9-bit signed offset is added to the contents of GPR base to form the effective address.
The SWE instruction functions the same as the SW instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment config- ured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Enhanced Virtual Addressing section for additional information.
Implementation of this instruction is specified by the Config5EVA field being set to 1.
Restrictions:
Only usable in kernel mode when accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode.
Pre-Release 6: The effective address must be naturally-aligned. If either of the 2 least-significant bits of the address is non-zero, an Address Error exception occurs.
Release 6 allows hardware to provide address misalignment support in lieu of requiring natural alignment.
Note: The pseudocode is not completely adapted for Release 6 misalignment support as the handling is implementa- tion dependent.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE) dataword  GPR[rt]
StoreMemory (CCA, WORD, dataword, pAddr, vAddr, DATA)
Exceptions:
TLB Refill, TLB Invalid, Bus Error, Address Error, Watch, Reserved Instruction, Coprocessor Unusable

SWL
31
26 25 21 20 16 15
655 16
Format: SWL rt, offset(base) Purpose: Store Word Left
To store the most-significant part of a word to an unaligned memory address.
Description: memory[GPR[base] + offset]  GPR[rt]
IStore Word Left 0
MIPS32, removed in Release 6
SWL 101010
base
rt
offset
The 16-bit signed offset is added to the contents of GPR base to form an effective address (EffAddr). EffAddr is the address of the most-significant of 4 consecutive bytes forming a word (W) in memory starting at an arbitrary byte boundary.
A part of W (the most-significant 1 to 4 bytes) is in the aligned word containing EffAddr. The same number of the most-significant (left) bytes from the word in GPR rt are stored into these bytes of W.
The following figure illustrates this operation using big-endian byte ordering for 32-bit and 64-bit registers. The four consecutive bytes in 2..5 form an unaligned word starting at location 2. A part of W (2 bytes) is located in the aligned word containing the most-significant byte at 2.
3. SWL stores the most-significant 2 bytes of the low word from the source register into these 2 bytes in memory.
4. The complementary SWR stores the remainder of the unaligned word.
Figure 5.9 Unaligned Word Store Using SWL and SWR
Word at byte 2 in memory, big-endian byte order; each memory byte contains its own address
most — significance — least
GPR 24
0
1
2
3
4
5
6
7
8

Memory: Initial contents
After executing SWL $24,2($0) Then after SWR $24,5($0)
E
F
G
H
0
1
E
F
4
5
6

0
1
E
F
G
H
6

The bytes stored from the source register to memory depend on both the offset of the effective address within an aligned word—that is, the low 2 bits of the address (vAddr1..0)—and the current byte-ordering mode of the processor (big- or little-endian). The following figure shows the bytes stored for every combination of offset and byte ordering.
The MIPS32® Instruction Set Manual, Revision 6.06 397

SWL
Store Word Left
Figure 5.10 Bytes Stored by an SWL Instruction
0
1
2
3 big-endian
offset (vAddr1. 0)
0 little-endian least
Memory contents and byte offsets
Initial contents of Dest Register
64-bit register
most — significance — least 32-bit register
i
j
k
l
A
B
C
D
E
F
G
H
3 most
1
2
— significance —
E
F
G
H
Memory contents after instruction (shaded is unchanged)
Big-endian byte ordering
vAddr1..0 0
1
2
3
Little-endian byte ordering
EFGH
i
EFG
i
j
EF
i
j
k
E
ijk
E
ij
EF
i
EFG
EFGH
398
The MIPS32® Instruction Set Manual, Revision 6.06
Restrictions:
None
Availability and Compatibility:
Release 6 removes the load/store-left/right family of instructions, and requires the system to support misaligned memory accesses.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE)
pAddrpAddr ||(pAddr xor ReverseEndian2) PSIZE-1..2 1..0
If BigEndianMem = 0 then
pAddr  pAddr || 02 PSIZE-1..2
endif
byte  vAddr xor BigEndianCPU2 1..0
dataword  024–8*byte || GPR[rt] 31..24-8*byte
StoreMemory(CCA, byte, dataword, pAddr, vAddr, DATA)
Exceptions:
TLB Refill, TLB Invalid, TLB Modified, Bus Error, Address Error, Watch

SWLE IStore Word Left EVA
31
26 25 21 20 16 15 7 6 5 0
655916
Format: SWLE rt, offset(base) MIPS32, removed in Release 6 Purpose: Store Word Left EVA
To store the most-significant part of a word to an unaligned user mode virtual address while operating in kernel mode.
Description: memory[GPR[base] + offset]  GPR[rt]
The 9-bit signed offset is added to the contents of GPR base to form an effective address (EffAddr). EffAddr is the address of the most-significant of 4 consecutive bytes forming a word (W) in memory starting at an arbitrary byte boundary.
A part of W (the most-significant 1 to 4 bytes) is in the aligned word containing EffAddr. The same number of the most-significant (left) bytes from the word in GPR rt are stored into these bytes of W.
The following figure shows this operation using big-endian byte ordering for 32-bit and 64-bit registers. The 4 con- secutive bytes in 2..5 form an unaligned word starting at location 2. A part of W (2 bytes) is located in the aligned word containing the most-significant byte at 2.
1. SWLE stores the most-significant 2 bytes of the low word from the source register into these 2 bytes in memory.
2. The complementary SWRE stores the remainder of the unaligned word.
Figure 5.11 Unaligned Word Store Using SWLE and SWRE
SPECIAL3 011111
base
rt
offset
0
SWLE 100001
Word at byte 2 in memory, big-endian byte order; each memory byte contains its own address
most — significance — least
GPR 24
0
1
2
3
4
5
6
7
8

Memory: Initial contents
After executing SWLE $24,2($0) Then after SWRE $24,5($0)
E
F
G
H
0
1
E
F
4
5
6

0
1
E
F
G
H
6

The bytes stored from the source register to memory depend on both the offset of the effective address within an aligned word—that is, the low 2 bits of the address (vAddr1..0)—and the current byte-ordering mode of the processor (big- or little-endian). The following figure shows the bytes stored for every combination of offset and byte ordering.
The SWLE instruction functions the same as the SWL instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment con- figured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Enhanced Virtual Addressing section for additional information.
Implementation of this instruction is specified by the Config5EVA field being set to 1.
The MIPS32® Instruction Set Manual, Revision 6.06 399

SWLE
Store Word Left EVA
Figure 5.12 Bytes Stored by an SWLE Instruction
0
1
2
3 big-endian
offset (vAddr1. 0)
0 little-endian least
Memory contents and byte offsets
Initial contents of Dest Register
64-bit register
most — significance — least 32-bit register
i
j
k
l
A
B
C
D
E
F
G
H
3 most
1
2
— significance —
E
F
G
H
Memory contents after instruction (shaded is unchanged)
Big-endian byte ordering
vAddr1..0 0
1
2
3
Little-endian byte ordering
EFGH
i
EFG
i
j
EF
i
j
k
E
ijk
E
ij
EF
i
EFG
EFGH
400
The MIPS32® Instruction Set Manual, Revision 6.06
Restrictions:
Only usable when access to Coprocessor0 is enabled and when accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode.
Availability and Compatibility:
Release 6 removes the load/store-left/right family of instructions, and requires the system to support misaligned memory accesses.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE)
pAddrpAddr ||(pAddr xor ReverseEndian2) PSIZE-1..2 1..0
If BigEndianMem = 0 then
pAddr  pAddr || 02 PSIZE-1..2
endif
byte  vAddr xor BigEndianCPU2 1..0
dataword  024–8*byte || GPR[rt] 31..24–8*byte
StoreMemory(CCA, byte, dataword, pAddr, vAddr, DATA)
Exceptions:
TLB Refill, TLB Invalid, TLB Modified, Bus Error, Address Error, Watch, Reserved Instruction, Coprocessor Unus- able

SWR
31
26 25 21 20 16 15
655 16
Format: SWR rt, offset(base) Purpose: Store Word Right
To store the least-significant part of a word to an unaligned memory address.
Description: memory[GPR[base] + offset]  GPR[rt]
IStore Word Right 0
MIPS32, removed in Release 6
SWR 101110
base
rt
offset
The 16-bit signed offset is added to the contents of GPR base to form an effective address (EffAddr). EffAddr is the address of the least-significant of 4 consecutive bytes forming a word (W) in memory starting at an arbitrary byte boundary.
A part of W (the least-significant 1 to 4 bytes) is in the aligned word containing EffAddr. The same number of the least-significant (right) bytes from the word in GPR rt are stored into these bytes of W.
The following figure illustrates this operation using big-endian byte ordering for 32-bit and 64-bit registers. The 4 consecutive bytes in 2..5 form an unaligned word starting at location 2. A part of W (2 bytes) is contained in the aligned word containing the least-significant byte at 5.
1. SWR stores the least-significant 2 bytes of the low word from the source register into these 2 bytes in memory.
2. The complementary SWL stores the remainder of the unaligned word.
Figure 5.13 Unaligned Word Store Using SWR and SWL
Word at byte 2 in memory, big-endian byte order, each mem byte contains its address
least — significance — least
GPR 24
0
1
2
3
4
5
6
7
8

Memory: Initial contents
After executing SWR $24,5($0) Then after SWL $24,2($0)
E
F
G
H
0
1
2
3
G
H
6

0
1
E
F
G
H
6

The bytes stored from the source register to memory depend on both the offset of the effective address within an aligned word—that is, the low 2 bits of the address (vAddr1..0)—and the current byte-ordering mode of the processor (big- or little-endian). The following figure shows the bytes stored for every combination of offset and byte-ordering.
The MIPS32® Instruction Set Manual, Revision 6.06 401

402
The MIPS32® Instruction Set Manual, Revision 6.06
SWR
Store Word Right
PSIZE-1..2 If BigEndianMem = 0 then
1..0
pAddr  pAddr endif
|| 02 byte  vAddr xor BigEndianCPU2
dataword  GPR[rt] Exceptions:
Figure 5.14 Bytes Stored by SWR Instruction
0
1
2
3 big-endian
offset (vAddr1. 0)
0 little-endian least
Memory contents and byte offsets
Initial contents of Dest Register
64-bit register
most — significance — least 32-bit register
i
j
k
l
A
B
C
D
E
F
G
H
3 most
1
2
— significance —
E
F
G
H
Memory contents after instruction (shaded is unchanged)
Big-endian byte ordering
vAddr1..0 0
1
2
3
Little-endian byte ordering
H
jkl
GH
kl
FGH
l
EFGH
EFGH
FGH
l
GH
k
l
H
j
k
l
Restrictions:
None
Availability and Compatibility:
Release 6 removes the load/store-left/right family of instructions, and requires the system to support misaligned memory accesses.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE) pAddr  pAddr || (pAddr xor ReverseEndian2)
PSIZE-1..2
1..0
|| 08*byte
StoreMemory(CCA, WORD-byte, dataword, pAddr, vAddr, DATA)
31–8*byte
TLB Refill, TLB Invalid, TLB Modified, Bus Error, Address Error, Watch

SWR IStore Word Right
The MIPS32® Instruction Set Manual, Revision 6.06 403

SWRE Store Word Right EVA
31
26 25 21 20 16 15 7 6 5 0
655916
Format: SWRE rt, offset(base) MIPS32, removed in Release 6 Purpose: Store Word Right EVA
To store the least-significant part of a word to an unaligned user mode virtual address while operating in kernel mode.
Description: memory[GPR[base] + offset]  GPR[rt]
The 9-bit signed offset is added to the contents of GPR base to form an effective address (EffAddr). EffAddr is the address of the least-significant of 4 consecutive bytes forming a word (W) in memory starting at an arbitrary byte boundary.
A part of W (the least-significant 1 to 4 bytes) is in the aligned word containing EffAddr. The same number of the least-significant (right) bytes from the word in GPR rt are stored into these bytes of W.
The following figure illustrates this operation using big-endian byte ordering for 32-bit and 64-bit registers. The 4 consecutive bytes in 2..5 form an unaligned word starting at location 2. A part of W (2 bytes) is contained in the aligned word containing the least-significant byte at 5.
3. SWRE stores the least-significant 2 bytes of the low word from the source register into these 2 bytes in memory.
4. The complementary SWLE stores the remainder of the unaligned word.
Figure 5.15 Unaligned Word Store Using SWRE and SWLE
SPECIAL3 011111
base
rt
offset
0
SWRE 100010
Word at byte 2 in memory, big-endian byte order, each mem byte contains its address
least — significance — least
GPR 24
0
1
2
3
4
5
6
7
8

Memory: Initial contents
After executing SWRE $24,5($0) Then after SWLE $24,2($0)
E
F
G
H
0
1
2
3
G
H
6

0
1
E
F
G
H
6

404
The bytes stored from the source register to memory depend on both the offset of the effective address within an aligned word—that is, the low 2 bits of the address (vAddr1..0)—and the current byte-ordering mode of the processor (big- or little-endian). The following figure shows the bytes stored for every combination of offset and byte-ordering.
The LWE instruction functions the same as the LW instruction, except that address translation is performed using the user mode virtual address space mapping in the TLB when accessing an address within a memory segment config- ured to use the MUSUK access mode. Memory segments using UUSK or MUSK access modes are also accessible. Refer to Volume III, Enhanced Virtual Addressing section for additional information.
Implementation of this instruction is specified by the Config5EVA field being set to 1.
The MIPS32® Instruction Set Manual, Revision 6.06

SWRE
IStore Word Right EVA
Figure 5.16 Bytes Stored by SWRE Instruction
0
1
2
3 big-endian
offset (vAddr1. 0)
0 little-endian least
Memory contents and byte offsets
Initial contents of Dest Register
64-bit register
most — significance — least 32-bit register
i
j
k
l
A
B
C
D
E
F
G
H
3 most
1
2
— significance —
E
F
G
H
Memory contents after instruction (shaded is unchanged)
Big-endian byte ordering
vAddr1..0 0
1
2
3
Little-endian byte ordering
H
jkl
GH
kl
FGH
l
EFGH
EFGH
FGH
l
GH
k
l
H
j
k
l
Restrictions:
Only usable when access to Coprocessor0 is enabled and when accessing an address within a segment configured using UUSK, MUSK or MUSUK access mode.
Availability and Compatibility:
Release 6 removes the load/store-left/right family of instructions, and requires the system to support misaligned memory accesses.
Operation:
vAddr  sign_extend(offset) + GPR[base]
(pAddr, CCA)  AddressTranslation (vAddr, DATA, STORE) pAddr  pAddr || (pAddr xor ReverseEndian2)
PSIZE-1..2 If BigEndianMem = 0 then
1..0
pAddr  pAddr endif
|| 02 byte  vAddr xor BigEndianCPU2
dataword  GPR[rt] Exceptions:
PSIZE-1..2
1..0
|| 08*byte
StoreMemory(CCA, WORD-byte, dataword, pAddr, vAddr, DATA)
31–8*byte
TLB Refill, TLB Invalid, TLB Modified, Bus Error, Address Error, Watch, Coprocessor Unusable
The MIPS32® Instruction Set Manual, Revision 6.06 405

SWXC1 Store Word Indexed from Floating Point
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: SWXC1 fs, index(base) MIPS64, MIPS32 Release 2, removed in Release 6 Purpose: Store Word Indexed from Floating Point
To store a word from an FPR to memory (GPR+GPR addressing)
Description: memory[GPR[base] + GPR[index]]  FPR[fs]
The low 32-bit word from FPR fs is stored in memory at the location specified by the aligned effective address. The
contents of GPR index and GPR base are added to form the effective address. Restrictions:
An Address Error exception occurs if EffectiveAddress1..0 ≠ 0 (not word-aligned).
Availability and Compatibility:
This instruction has been removed in Release 6.
Required in all versions of MIPS64 since MIPS64 Release 1. Not available in MIPS32 Release 1. Required in MIPS32 Release 2 and all subsequent versions of MIPS32. When required, required whenever FPU is present, whether a 32-bit or 64-bit FPU, whether in 32-bit or 64-bit FP Register Mode (FIRF64=0 or 1, StatusFR=0 or 1).
Operation:
vAddr  GPR[base] + GPR[index]
COP1X 010011
base
index
fs
0 00000
SWXC1 001000
406
The MIPS32® Instruction Set Manual, Revision 6.06
 03 then SignalException(AddressError)
endif
(pAddr, CCA)  AddressTranslation(vAddr, DATA, STORE) dataword  ValueFPR(fs, UNINTERPRETED_WORD) StoreMemory(CCA, WORD, dataword, pAddr, vAddr, DATA)
Exceptions:
TLB Refill, TLB Invalid, TLB Modified, Address Error, Reserved Instruction, Coprocessor Unusable, Watch
if vAddr
1..0

SYNC ISynchronize Shared Memory
31
26 25 21 20 16 15 11 10 6 5
6 15 5 6
Format: SYNC (stype = 0 implied) SYNC stype
Purpose: Synchronize Shared Memory
To order loads and stores for shared memory.
Release 6 (with Config5GI =10/11) extends SYNC for Global Invalidate instructions (GINVI/GINVT).
Description:
These types of ordering guarantees are available through the SYNC instruction:
• Completion Barriers
• Ordering Barriers
Completion Barrier — Simple Description:
0
MIPS32 MIPS32
SPECIAL 000000
0
00 0000 0000 0000 0
stype
SYNC 001111
• The barrier affects only uncached and cached coherent loads and stores.
• The specified memory instructions (loads or stores or both) that occur before the SYNC instruction must be
completed before the specified memory instructions after the SYNC are allowed to start.
• Loads are completed when the destination register is written. Stores are completed when the stored value is visible to every other processor in the system.
Completion Barrier — Detailed Description:
• Every synchronizable specified memory instruction (loads or stores or both) that occurs in the instruction stream before the SYNC instruction must be already globally performed before any synchronizable speci- fied memory instructions that occur after the SYNC are allowed to be performed, with respect to any other processor or coherent I/O module.
• The barrier does not guarantee the order in which instruction fetches are performed.
• A stype value of zero will always be defined such that it performs the most complete set of synchronization operations that are defined.This means stype zero always does a completion barrier that affects both loads and stores preceding the SYNC instruction and both loads and stores that are subsequent to the SYNC instruction. Non-zero values of stype may be defined by the architecture or specific implementations to per- form synchronization behaviors that are less complete than that of stype zero. If an implementation does not use one of these non-zero values to define a different synchronization behavior, then that non-zero value of stype must act the same as stype zero completion barrier. This allows software written for an implementation with a lighter-weight barrier to work on another implementation which only implements the stype zero com- pletion barrier.
• A completion barrier is required, potentially in conjunction with SSNOP (in Release 1 of the Architecture) or EHB (in Release 2 of the Architecture), to guarantee that memory reference results are visible across operating mode changes. For example, a completion barrier is required on some implementations on entry to and exit from Debug Mode to guarantee that memory effects are handled correctly.
The MIPS32® Instruction Set Manual, Revision 6.06 407

408
The MIPS32® Instruction Set Manual, Revision 6.06
SYNC Synchronize Shared Memory SYNC behavior when the stype field is zero:
• A completion barrier that affects preceding loads and stores and subsequent loads and stores.
Ordering Barrier — Simple Description:
• The barrier affects only uncached and cached coherent loads and stores.
• The specified memory instructions (loads or stores or both) that occur before the SYNC instruction must always be ordered before the specified memory instructions after the SYNC.
• Memory instructions which are ordered before other memory instructions are processed by the load/store datapath first before the other memory instructions.
Ordering Barrier — Detailed Description:
• Every synchronizable specified memory instruction (loads or stores or both) that occurs in the instruction stream before the SYNC instruction must reach a stage in the load/store datapath after which no instruction re-ordering is possible before any synchronizable specified memory instruction which occurs after the SYNC instruction in the instruction stream reaches the same stage in the load/store datapath.
• If any memory instruction before the SYNC instruction in program order, generates a memory request to the external memory and any memory instruction after the SYNC instruction in program order also generates a memory request to external memory, the memory request belonging to the older instruction must be globally performed before the time the memory request belonging to the younger instruction is globally performed.
• The barrier does not guarantee the order in which instruction fetches are performed.
As compared to the completion barrier, the ordering barrier is a lighter-weight operation as it does not require the specified instructions before the SYNC to be already completed. Instead it only requires that those specified instruc- tions which are subsequent to the SYNC in the instruction stream are never re-ordered for processing ahead of the specified instructions which are before the SYNC in the instruction stream. This potentially reduces how many cycles the barrier instruction must stall before it completes.
The Acquire and Release barrier types are used to minimize the memory orderings that must be maintained and still have software synchronization work.
Implementations that do not use any of the non-zero values of stype to define different barriers, such as ordering bar- riers, must make those stype values act the same as stype zero.
For the purposes of this description, the CACHE, PREF and PREFX instructions are treated as loads and stores. That is, these instructions and the memory transactions sourced by these instructions obey the ordering and completion rules of the SYNC instruction.
If Global Invalidate instructions are supported in Release 6, then SYNC (stype=0x14) acts as a completion barrier to ensure completion of any preceding GINVI or GINVT operation. This SYNC operation is globalized and only com- pletes if all preceding GINVI or GINVT operations related to the same program have completed in the system. (Any references to GINVT also imply GINVGT, available in a virtualized MIPS system. GINVT however will be used exclusively.)
A system that implements the Global Invalidates also requires that the completion of this SYNC be constrained by legacy SYNCI operations. Thus SYNC (stype=0x14) can also be used to determine whether preceding (in program order) SYNCI operations have completed.

SYNC ISynchronize Shared Memory The SYNC (stype=0x14) also act as an ordering barrier as described in Table 5.5.
In the typical use cases, a single GINVI is used by itself to invalidate caches and would be followed by a SYNC (stype=0x14).
In the case of GINVT, multiple GINVT could be used to invalidate multiple TLB mappings, and the SYNC (stype=0x14) would be used to guaranteed completion of any number of GINVTs preceding it.
Table 5.5 lists the available completion barrier and ordering barriers behaviors that can be specified using the stype field.
Table 5.5 Encodings of the Bits[10:6] of the SYNC instruction; the SType Field
Code
Name
Older instructions which must reach the load/store ordering point before the SYNC instruction completes.
Younger instructions which must reach the load/store ordering point only after the SYNC instruction completes.
Older instructions which must be globally performed when the SYNC instruction completes
Compliance
0x0
SYNC or SYNC 0
Loads, Stores
Loads, Stores
Loads, Stores
Required
0x4
SYNC_WMB or SYNC 4
Stores
Stores
Optional
0x10
SYNC_MB or SYNC 16
Loads, Stores
Loads, Stores
Optional
0x11
SYNC_ACQUIRE or
SYNC 17
Loads
Loads, Stores
Optional
0x12
SYNC_RELEASE or
SYNC 18
Loads, Stores
Stores
Optional
0x13
SYNC_RMB or SYNC 19
Loads
Loads
Optional
0x1-0x3, 0x5-0xF
Implementation-Spe- cific and Vendor Spe- cific Sync Types
0x14
SYNC_GINV
Loads, Stores
Loads, Stores
GINVI, GINVT, SYNCI
Release 6 w/ Config5GI =10/11 otherwise Reserved
0x15 – 0x1F
RESERVED
Reserved for MIPS Technologies for future extension of the architecture.
The MIPS32® Instruction Set Manual, Revision 6.06 409

410
The MIPS32® Instruction Set Manual, Revision 6.06
SYNC Synchronize Shared Memory Terms:
Synchronizable: A load or store instruction is synchronizable if the load or store occurs to a physical location in shared memory using a virtual location with a memory access type of either uncached or cached coherent. Shared memory is memory that can be accessed by more than one processor or by a coherent I/O system module.
Performed load: A load instruction is performed when the value returned by the load has been determined. The result of a load on processor A has been determined with respect to processor or coherent I/O module B when a subsequent store to the location by B cannot affect the value returned by the load. The store by B must use the same memory access type as the load.
Performed store: A store instruction is performed when the store is observable. A store on processor A is observable with respect to processor or coherent I/O module B when a subsequent load of the location by B returns the value written by the store. The load by B must use the same memory access type as the store.
Globally performed load: A load instruction is globally performed when it is performed with respect to all processors and coherent I/O modules capable of storing to the location.
Globally performed store: A store instruction is globally performed when it is globally observable. It is globally observable when it is observable by all processors and I/O modules capable of loading from the location.
Coherent I/O module: A coherent I/O module is an Input/Output system component that performs coherent Direct Memory Access (DMA). It reads and writes memory independently as though it were a processor doing loads and stores to locations with a memory access type of cached coherent.
Load/Store Datapath: The portion of the processor which handles the load/store data requests coming from the pro- cessor pipeline and processes those requests within the cache and memory system hierarchy.
Restrictions:
The effect of SYNC on the global order of loads and stores for memory access types other than uncached and cached coherent is UNPREDICTABLE.
Operation:
SyncOperation(stype)
Exceptions:
None
Programming Notes:
A processor executing load and store instructions observes the order in which loads and stores using the same mem- ory access type occur in the instruction stream; this is known as program order.
A parallel program has multiple instruction streams that can execute simultaneously on different processors. In mul- tiprocessor (MP) systems, the order in which the effects of loads and stores are observed by other processors—the global order of the loads and store—determines the actions necessary to reliably share data in parallel programs.
When all processors observe the effects of loads and stores in program order, the system is strongly ordered. On such systems, parallel programs can reliably share data without explicit actions in the programs. For such a system, SYNC has the same effect as a NOP. Executing SYNC on such a system is not necessary, but neither is it an error.
If a multiprocessor system is not strongly ordered, the effects of load and store instructions executed by one processor may be observed out of program order by other processors. On such systems, parallel programs must take explicit actions to reliably share data. At critical points in the program, the effects of loads and stores from an instruction stream must occur in the same order for all processors. SYNC separates the loads and stores executed on the proces- sor into two groups, and the effect of all loads and stores in one group is seen by all processors before the effect of any load or store in the subsequent group. In effect, SYNC causes the system to be strongly ordered for the executing processor at the instant that the SYNC is executed.

SYNC ISynchronize Shared Memory
Many MIPS-based multiprocessor systems are strongly ordered or have a mode in which they operate as strongly ordered for at least one memory access type. The MIPS architecture also permits implementation of MP systems that are not strongly ordered; SYNC enables the reliable use of shared memory on such systems. A parallel program that does not use SYNC generally does not operate on a system that is not strongly ordered. However, a program that does use SYNC works on both types of systems. (System-specific documentation describes the actions needed to reliably share data in parallel programs for that system.)
The behavior of a load or store using one memory access type is UNPREDICTABLE if a load or store was previ- ously made to the same physical location using a different memory access type. The presence of a SYNC between the references does not alter this behavior.
SYNC affects the order in which the effects of load and store instructions appear to all processors; it does not gener- ally affect the physical memory-system ordering or synchronization issues that arise in system programming. The effect of SYNC on implementation-specific aspects of the cached memory system, such as writeback buffers, is not defined.
# Processor A (writer)
# Conditions at entry:
# The value 0 has been stored in FLAG and that value is observable by B
SW R1, DATA LI R2,1 SYNC
SW
# change shared DATA value
# Perform DATA store before performing FLAG store
# say that the shared DATA value is valid
R2, FLAG
# Processor B (reader) LI R2, 1
1: LW
BNE R2, R1, 1B# if it says that DATA is not valid, poll again NOP
SYNC # FLAG value checked before doing DATA read
LW R1, DATA # Read (valid) shared DATA value
The code fragments above shows how SYNC can be used to coordinate the use of shared data between separate writer and reader instruction streams in a multiprocessor environment. The FLAG location is used by the instruction streams to determine whether the shared data item DATA is valid. The SYNC executed by processor A forces the store of DATA to be performed globally before the store to FLAG is performed. The SYNC executed by processor B ensures that DATA is not read until after the FLAG value indicates that the shared data is valid.
R1, FLAG # Get FLAG
The MIPS32® Instruction Set Manual, Revision 6.06 411

SYNCI Synchronize Caches to Make Instruction Writes Effective
Release 6
31
26 25 21 20 16 15
655 16
Format: SYNCI offset(base)
Purpose: Synchronize Caches to Make Instruction Writes Effective
To synchronize all caches to make instruction writes effective.
Description:
0
MIPS32 Release 2
REGIMM 000001
base
SYNCI 11111
offset
412
The MIPS32® Instruction Set Manual, Revision 6.06
This instruction is used after a new instruction stream is written to make the new instructions effective relative to an instruction fetch, when used in conjunction with the SYNC and JALR.HB, JR.HB, or ERET instructions, as described below. Unlike the CACHE instruction, the SYNCI instruction is available in all operating modes in an implementation of Release 2 of the architecture.
The 16-bit offset is sign-extended and added to the contents of the base register to form an effective address. The effective address is used to address the cache line in all caches which may need to be synchronized with the write of the new instructions. The operation occurs only on the cache line which may contain the effective address. One SYNCI instruction is required for every cache line that was written. See the Programming Notes below.
A TLB Refill and TLB Invalid (both with cause code equal TLBL) exception can occur as a by product of this instruction. This instruction never causes TLB Modified exceptions nor TLB Refill exceptions with a cause code of TLBS. This instruction never causes Execute-Inhibit nor Read-Inhibit exceptions.
A Cache Error exception may occur as a by product of this instruction. For example, if a writeback operation detects a cache or bus error during the processing of the operation, that error is reported via a Cache Error exception. Simi- larly, a Bus Error Exception may occur if a bus operation invoked by this instruction is terminated in an error.
An Address Error Exception (with cause code equal AdEL) may occur if the effective address references a portion of the kernel address space which would normally result in such an exception. It is implementation dependent whether such an exception does occur.
It is implementation dependent whether a data watch is triggered by a SYNCI instruction whose address matches the Watch register address match conditions.
Restrictions:
The operation of the processor is UNPREDICTABLE if the effective address references any instruction cache line that contains instructions to be executed between the SYNCI and the subsequent JALR.HB, JR.HB, or ERET instruc- tion required to clear the instruction hazard.
The SYNCI instruction has no effect on cache lines that were previously locked with the CACHE instruction. If cor- rect software operation depends on the state of a locked line, the CACHE instruction must be used to synchronize the caches.
Full visibility of the new instruction stream requires execution of a subsequent SYNC instruction, followed by a JALR.HB, JR.HB, DERET, or ERET instruction. The operation of the processor is UNPREDICTABLE if this sequence is not followed.
SYNCI globalization:
The SYNCI instruction acts on the current processor at a minimum. Implementations are required to affect caches outside the current processor to perform the operation on the current processor (as might be the case if multiple pro- cessors share an L2 or L3 cache).

SYNCI ISynchronize Caches to Make Instruction Writes Effective In multiprocessor implementations where instruction caches are coherently maintained by hardware, the SYNCI
instruction should behave as a NOP instruction.
In multiprocessor implementations where instruction caches are not coherently maintained by hardware, the SYNCI instruction may optionally affect all coherent icaches within the system. If the effective address uses a coherent Cacheability and Coherency Attribute (CCA), then the operation may be globalized, meaning it is broadcast to all of the coherent instruction caches within the system. If the effective address does not use one of the coherent CCAs, there is no broadcast of the SYNCI operation. If multiple levels of caches are to be affected by one SYNCI instruc- tion, all of the affected cache levels must be processed in the same manner – either all affected cache levels use the globalized behavior or all affected cache levels use the non-globalized behavior.
Pre-Release 6: Portable software could not rely on the optional globalization of SYNCI. Strictly portable software without implementation specific awareness could only rely on expensive “instruction cache shootdown” using inter- processor interrupts.
Release 6: SYNCI globalization is required. Compliant implementations must globalize SYNCI, and portable soft- ware can rely on this behavior.
Operation:
vaddr  GPR[base] + sign_extend(offset) SynchronizeCacheLines(vaddr) /* Operate on all caches */
Exceptions:
Reserved Instruction exception (Release 1 implementations only) TLB Refill Exception
TLB Invalid Exception
Address Error Exception
Cache Error Exception Bus Error Exception
Programming Notes:
When the instruction stream is written, the SYNCI instruction should be used in conjunction with other instructions to make the newly-written instructions effective. The following example shows a routine which can be called after the new instruction stream is written to make those changes effective. The SYNCI instruction could be replaced with the corresponding sequence of CACHE instructions (when access to Coprocessor 0 is available), and that the JR.HB instruction could be replaced with JALR.HB, ERET, or DERET instructions, as appropriate. A SYNC instruction is required between the final SYNCI instruction in the loop and the instruction that clears instruction hazards.
/*
* This routine makes changes to the instruction stream effective to the
* hardware. It should be called after the instruction stream is written.
* On return, the new instructions are effective.
*
* Inputs:
* * */
a0 = Start address of new instruction stream
a1 = Size, in bytes, of new instruction stream
/* If size==0, */
/* branch around */
/* Calculate end address + 1 */
/* Get step size for SYNCI from new */
/* Release 2 instruction */
/* If no caches require synchronization, */
/* branch around */
beq
nop
addu
rdhwr v0, HWSYNCIStep
a1, zero, 20f
a1, a0, a1
beq v0, zero, 20f nop
The MIPS32® Instruction Set Manual, Revision 6.06
413

414
The MIPS32® Instruction Set Manual, Revision 6.06
SYNCI
Synchronize Caches to Make Instruction Writes Effective
10: synci addu
sltu
bne
nop
sync
20: jr.hb nop
0(a0)
a0, a0, v0
v1, a0, a1
v1, zero, 10b
ra
/*
/*
/*
/*
/*
/*
/*
Synchronize all caches around address */
Add step size in delay slot */
Compare current with end address */
Branch if more to do */
branch around */
Clear memory hazards */
Return, clearing instruction hazards */

SYSCALL ISystem Call
31
26 25 6 5 0
6 20 6
Format: SYSCALL MIPS32 Purpose: System Call
To cause a System Call exception.
Description:
A system call exception occurs, immediately and unconditionally transferring control to the exception handler.
The code field is available for use as software parameters, but may be retrieved by the exception handler by loading the contents of the memory word containing the instruction. Alternatively, if CP0 BadInstr is implemented, the code field may be obtained from BadInstr.
Restrictions:
None
Operation:
SignalException(SystemCall)
Exceptions:
System Call
SPECIAL 000000
code
SYSCALL 001100
The MIPS32® Instruction Set Manual, Revision 6.06 415

TEQ
Trap if Equal
31
26 25 21 20 16 15 6 5 0
6 5 5 10 6
Format: TEQ rs, rt MIPS32 Purpose: Trap if Equal
To compare GPRs and do a conditional trap.
Description: if GPR[rs] = GPR[rt] then Trap
Compare the contents of GPR rs and GPR rt as signed integers. If GPR rs is equal to GPR rt, then take a Trap excep-
tion.
The contents of the code field are ignored by hardware and may be used to encode information for system software. To retrieve the information, system software may load the instruction word from memory. Alternatively, if CP0 BadInstr is implemented, the code field may be obtained from BadInstr.
Restrictions:
None
Operation:
if GPR[rs] = GPR[rt] then
SignalException(Trap)
endif
Exceptions:
Trap
SPECIAL 000000
rs
rt
code
TEQ 110100
416
The MIPS32® Instruction Set Manual, Revision 6.06

TEQI
31
26 25 21 20 16 15
655 16
Format: TEQI rs, immediate Purpose: Trap if Equal Immediate
To compare a GPR to a constant and do a conditional trap.
ITrap if Equal Immediate 0
MIPS32, removed in Release 6
REGIMM 000001
rs
TEQI 01100
immediate
Description: if GPR[rs] = immediate then Trap
Compare the contents of GPR rs and the 16-bit signed immediate as signed integers. If GPR rs is equal to immediate,
then take a Trap exception.
Restrictions:
None
Availability and Compatibility:
This instruction has been removed in Release 6.
Operation:
if GPR[rs] = sign_extend(immediate) then
SignalException(Trap)
endif
Exceptions:
Trap
The MIPS32® Instruction Set Manual, Revision 6.06 417

TGE
Trap if Greater or Equal
31
26 25 21 20 16 15 6 5 0
6 5 5 10 6
Format: TGE rs, rt MIPS32 Purpose: Trap if Greater or Equal
To compare GPRs and do a conditional trap.
Description: if GPR[rs]  GPR[rt] then Trap
Compare the contents of GPR rs and GPR rt as signed integers. If GPR rs is greater than or equal to GPR rt, then take
a Trap exception.
The contents of the code field are ignored by hardware and may be used to encode information for system software. To retrieve the information, the system software may load the instruction word from memory. Alternatively, if CP0 BadInstr is implemented, the code field may be obtained from BadInstr.
Restrictions:
None
Operation:
if GPR[rs]  GPR[rt] then SignalException(Trap)
endif
Exceptions:
Trap
SPECIAL 000000
rs
rt
code
TGE 110000
418
The MIPS32® Instruction Set Manual, Revision 6.06

TGEI
31
ITrap if Greater or Equal Immediate 0
26 25 21 20 16 15
655 16
Format: TGEI rs, immediate Purpose: Trap if Greater or Equal Immediate
To compare a GPR to a constant and do a conditional trap.
MIPS32, removed in Release 6
REGIMM 000001
rs
TGEI 01000
immediate
Description: if GPR[rs]  immediate then Trap
Compare the contents of GPR rs and the 16-bit signed immediate as signed integers. If GPR rs is greater than or equal
to immediate, then take a Trap exception. Restrictions:
None
Availability and Compatibility:
This instruction has been removed in Release 6.
Operation:
if GPR[rs]  sign_extend(immediate) then SignalException(Trap)
endif
Exceptions:
Trap
The MIPS32® Instruction Set Manual, Revision 6.06 419

TGEIU Trap if Greater or Equal Immediate Unsigned
31
26 25 21 20 16 15
655 16
Format: TGEIU rs, immediate
Purpose: Trap if Greater or Equal Immediate Unsigned
To compare a GPR to a constant and do a conditional trap.
0
MIPS32, removed in Release 6
REGIMM 000001
rs
TGEIU 01001
immediate
420
The MIPS32® Instruction Set Manual, Revision 6.06
Description: if GPR[rs]  immediate then Trap
Compare the contents of GPR rs and the 16-bit sign-extended immediate as unsigned integers. If GPR rs is greater
than or equal to immediate, then take a Trap exception.
Because the 16-bit immediate is sign-extended before comparison, the instruction can represent the smallest or largest unsigned numbers. The representable values are at the minimum [0, 32767] or maximum [max_unsigned-32767, max_unsigned] end of the unsigned range.
Restrictions:
None
Availability and Compatibility:
This instruction has been removed in Release 6.
Operation:
if (0 || GPR[rs])  (0 || sign_extend(immediate)) then SignalException(Trap)
endif
Exceptions:
Trap

TGEU ITrap if Greater or Equal Unsigned
31
26 25 21 20 16 15 6 5 0
6 5 5 10 6
Format: TGEU rs, rt MIPS32 Purpose: Trap if Greater or Equal Unsigned
To compare GPRs and do a conditional trap.
Description: if GPR[rs]  GPR[rt] then Trap
Compare the contents of GPR rs and GPR rt as unsigned integers. If GPR rs is greater than or equal to GPR rt, then
take a Trap exception.
The contents of the code field are ignored by hardware and may be used to encode information for system software. To retrieve the information, the system software may load the instruction word from memory. Alternatively, if CP0 BadInstr is implemented, the code field may be obtained from BadInstr.
Restrictions:
None
Operation:
if (0 || GPR[rs])  (0 || GPR[rt]) then SignalException(Trap)
endif
Exceptions:
Trap
SPECIAL 000000
rs
rt
code
TGEU 110001
The MIPS32® Instruction Set Manual, Revision 6.06 421

TLBINV TLB Invalidate
31
26 25 24 6 5 0
61 19 6
Format: TLBINV MIPS32 Purpose: TLB Invalidate
TLBINV invalidates a set of TLB entries based on ASID and Index match. The virtual address is ignored in the entry match. TLB entries which have their G bit set to 1 are not modified.
Implementation of the TLBINV instruction is optional. The implementation of this instruction is indicated by the IE field in Config4.
Support for TLBINV is recommend for implementations supporting VTLB/FTLB type of MMU. Implementation of EntryHIEHINV field is required for implementation of TLBINV instruction.
Description:
On execution of the TLBINV instruction, the set of TLB entries with matching ASID are marked invalid, excluding those TLB entries which have their G bit set to 1.
The EntryHIASID field has to be set to the appropriate ASID value before executing the TLBINV instruction. Behavior of the TLBINV instruction applies to all applicable TLB entries and is unaffected by the setting of the Wired
register.
• For JTLB-based MMU (ConfigMT=1):
All matching entries in the JTLB are invalidated. The Index register is unused.
• For VTLB/FTLB -based MMU (ConfigMT=4):
If TLB invalidate walk is implemented in software (Config4IE=2), then software must do these steps to flush the entire MMU:
1. one TLBINV instruction is executed with an index in VTLB range (invalidates all matching VTLB entries)
2. a TLBINV instruction is executed for each FTLB set (invalidates all matching entries in FTLB set)
If TLB invalidate walk is implemented in hardware (Config4IE=3), then software must do these steps to flush the entire MMU:
1. one TLBINV instruction is executed (invalidates all matching entries in both FTLB & VTLB). In this case, Index is unused.
Restrictions:
When Config4MT = 4 and Config4IE = 2, the operation is UNDEFINED if the contents of the Index register are
greater than or equal to the number of available TLB entries.
If access to Coprocessor 0 is not enabled, a Coprocessor Unusable Exception is signaled.
Availability and Compatibility:
Implementation of the TLBINV instruction is optional. The implementation of this instruction is indicated by the IE
COP0 010000
CO 1
0
000 0000 0000 0000 0000
TLBINV 000011
422
The MIPS32® Instruction Set Manual, Revision 6.06

TLBINV ITLB Invalidate
field in Config4.
Implementation of EntryHIEHINV field is required for implementation of TLBINV instruction.
Pre-Release 6, support for TLBINV is recommended for implementations supporting VTLB/FTLB type of MMU. Release 6 (and subsequent releases) support for TLBINV is required for implementations supporting VTLB/FTLB type of MMU.
Release 6: On processors that include a Block Address Translation (BAT) or Fixed Mapping (FM) MMU (ConfigMT = 2 or 3), the operation of this instruction causes a Reserved Instruction exception (RI).
Operation:
if ( ConfigMT=1 or (ConfigMT=4 & Config4IE=2 & Index < VTLBsize() )) startnum  0 endnum  VTLBsize() - 1 endif // treating VTLB and FTLB as one array if (ConfigMT=4 & Config4IE=2 & Index ≥ VTLBsize(); ) startnum  start of selected FTLB set // implementation specific endnum  end of selected FTLB set - 1 //implementation specifc endif if (ConfigMT=4 & Config4IE=3)) startnum  0 endnum  VTLBsize() + FTLBsize() - 1; endif for (i = startnum to endnum) if (TLB[i]ASID = EntryHiASID & TLB[i]G = 0) TLB[i]VPN2_invalid  1 endif endfor function VTLBsize SizeExt = ArchRev() ≥ 6 : Config4MMUExtDef ? Config4VTLBSizeExt ? Config4VTLBSizeExt ? Config4MMUSizeExt ; return 1 + ( (SizeExt << 6) | Config1.MMUSize ); endfunction function FTLBsize if ( Config1MT == 4 ) then return ( Config4FTLBWays + 2 ) * ( 1 << C0_Config4FTLBSets ); else return 0; endif endfunction Exceptions: Coprocessor Unusable, == 1 :0 : Config4MMUExtDef == 3 The MIPS32® Instruction Set Manual, Revision 6.06 423 TLBINV TLB Invalidate 424 The MIPS32® Instruction Set Manual, Revision 6.06 TLBINVF ITLB Invalidate Flush 31 26 25 24 6 5 0 61 19 6 Format: TLBINVF MIPS32 Purpose: TLB Invalidate Flush TLBINVF invalidates a set of TLB entries based on Index match. The virtual address and ASID are ignored in the entry match. Implementation of the TLBINVF instruction is optional. The implementation of this instruction is indicated by the IE field in Config4. Support for TLBINVF is recommend for implementations supporting VTLB/FTLB type of MMU. Implementation of the EntryHIEHINV field is required for implementation of TLBINV and TLBINVF instructions. Description: On execution of the TLBINVF instruction, all entries within range of Index are invalidated. Behavior of the TLBINVF instruction applies to all applicable TLB entries and is unaffected by the setting of the Wired register. • For JTLB-based MMU (ConfigMT=1): TLBINVF causes all entries in the JTLB to be invalidated. Index is unused. • For VTLB/FTLB-based MMU (ConfigMT=4): If TLB invalidate walk is implemented in your software (Config4IE=2), then your software must do these steps to flush the entire MMU: 1. one TLBINVF instruction is executed with an index in VTLB range (invalidates all VTLB entries) 2. a TLBINVF instruction is executed for each FTLB set (invalidates all entries in FTLB set) If TLB invalidate walk is implemented in hardware (Config4IE=3), then software must do these steps to flush the entire MMU: 1. one TLBINVF instruction is executed (invalidates all entries in both FTLB & VTLB). In this case, Index is unused. Restrictions: When ConfigMT=4 and ConfigIE=2, the operation is UNDEFINED if the contents of the Index register are greater than or equal to the number of available TLB entries. If access to Coprocessor 0 is not enabled, a Coprocessor Unusable Exception is signaled. Availability and Compatibility: Implementation of the TLBINVF instruction is optional. The implementation of this instruction is indicated by the IE field in Config4. COP0 010000 CO 1 0 000 0000 0000 0000 0000 TLBINVF 000100 Implementation of EntryHIEHINV field is required for implementation of TLBINVF instruction. The MIPS32® Instruction Set Manual, Revision 6.06 425 426 The MIPS32® Instruction Set Manual, Revision 6.06 TLBINVF TLB Invalidate Flush Pre-Release 6, support for TLBINVF is recommended for implementations supporting VTLB/FTLB type of MMU. Release 6 (and subsequent releases) support for TLBINV is required for implementations supporting VTLB/FTLB type of MMU. Release 6: On processors that include a Block Address Translation (BAT) or Fixed Mapping (FM) MMU (ConfigMT = 2 or 3), the operation of this instruction causes a Reserved Instruction exception (RI). Operation: if ( ConfigMT=1 or (ConfigMT=4 & Config4IE=2 & Index < VTLBsize() )) startnum  0 endnum  VTLBsize() - 1 endif // treating VTLB and FTLB as one array if (ConfigMT=4 & Config4IE=2 & Index ≥ VTLBsize(); ) startnum  start of selected FTLB set // implementation specific endnum  end of selected FTLB set - 1 //implementation specifc endif if (ConfigMT=4 & Config4IE=3)) startnum  0 endnum  TLBsize() + FTLBsize() - 1; endif for (i = startnum to endnum) TLB[i]VPN2_invalid  1 endfor function VTLBsize SizeExt = ArchRev() ≥ 6 : Config4MMUExtDef == 3 : Config4MMUExtDef == 1 :0 ; ? Config4VTLBSizeExt ? Config4VTLBSizeExt ? Config4MMUSizeExt return 1 + ( (SizeExt << 6) | Config1.MMUSize ); endfunction function FTLBsize if ( Config1MT == 4 ) then return ( Config4FTLBWays + 2 ) * ( 1 << C0_Config4FTLBSets ); else return 0; endif endfunction Exceptions: Coprocessor Unusable, TLBP IProbe TLB for Matching Entry 31 26 25 24 6 5 0 COP0 010000 CO 1 0 000 0000 0000 0000 0000 TLBP 001000 61 19 6 Format: TLBP Purpose: Probe TLB for Matching Entry To find a matching entry in the TLB. Description: MIPS32 The Index register is loaded with the address of the TLB entry whose contents match the contents of the EntryHi reg- ister. If no TLB entry matches, the high-order bit of the Index register is set. • In Release 1 of the Architecture, it is implementation dependent whether multiple TLB matches are detected on a TLBP. However, implementations are strongly encouraged to report multiple TLB matches only on a TLB write. • In Release 2 of the Architecture, multiple TLB matches may only be reported on a TLB write. • In Release 3 of the Architecture, multiple TLB matches may be reported on either TLB write or TLB probe. Restrictions: If access to Coprocessor 0 is not enabled, a Coprocessor Unusable Exception is signaled. Release 6: Processors that include a Block Address Translation (BAT) or Fixed Mapping (FM) MMU (ConfigMT = 2 or 3), the operation of this instruction causes a Reserved Instruction exception (RI). Operation: Index  1 || UNPREDICTABLE31 for i in 00 ... TLBEntries-1 if ((TLB[i]VPN2 and not (TLB[i]Mask)) = (EntryHiVPN2 and not (TLB[i]Mask))) and ((TLB[i]G = 1) or (TLB[i]ASID = EntryHiASID))then Index  i endif endfor Exceptions: Coprocessor Unusable, Machine Check The MIPS32® Instruction Set Manual, Revision 6.06 427 TLBR Read Indexed TLB Entry 31 26 25 24 6 5 0 COP0 010000 CO 1 0 000 0000 0000 0000 0000 TLBR 000001 428 The MIPS32® Instruction Set Manual, Revision 6.06 61 19 6 Format: TLBR Purpose: Read Indexed TLB Entry To read an entry from the TLB. Description: MIPS32 The EntryHi, EntryLo0, EntryLo1, and PageMask registers are loaded with the contents of the TLB entry pointed to by the Index register. • In Release 1 of the Architecture, it is implementation dependent whether multiple TLB matches are detected on a TLBR. However, implementations are strongly encouraged to report multiple TLB matches only on a TLB write. • In Release 2 of the Architecture, multiple TLB matches may only be reported on a TLB write. • In Release 3 of the Architecture, multiple TLB matches may be detected on a TLBR. In an implementation supporting TLB entry invalidation (Config4IE ≥ 1), reading an invalidated TLB entry causes EntryLo0 and EntryLo1 to be set to 0, EntryHiEHINV to be set to 1, all other EntryHi bits to be set to 0, and PageMask to be set to a value representing the minimum supported page size.. The value written to the EntryHi, EntryLo0, and EntryLo1 registers may be different from the original written value to the TLB via these registers in that: • The value returned in the VPN2 field of the EntryHi register may have those bits set to zero corresponding to the one bits in the Mask field of the TLB entry (the least-significant bit of VPN2 corresponds to the least-significant bit of the Mask field). It is implementation dependent whether these bits are preserved or zeroed after a TLB entry is written and then read. • The value returned in the PFN field of the EntryLo0 and EntryLo1 registers may have those bits set to zero cor- responding to the one bits in the Mask field of the TLB entry (the least significant bit of PFN corresponds to the least significant bit of the Mask field). It is implementation dependent whether these bits are preserved or zeroed after a TLB entry is written and then read. • The value returned in the G bit in both the EntryLo0 and EntryLo1 registers comes from the single G bit in the TLB entry. Recall that this bit was set from the logical AND of the two G bits in EntryLo0 and EntryLo1 when the TLB was written. Restrictions: The operation is UNDEFINED if the contents of the Index register are greater than or equal to the number of TLB entries in the processor. If access to Coprocessor 0 is not enabled, a Coprocessor Unusable Exception is signaled. Release 6: Processors that include a Block Address Translation (BAT) or Fixed Mapping (FM) MMU (ConfigMT = 2 or 3), the operation of this instruction causes a Reserved Instruction exception (RI). Operation: i  Index if i > (TLBEntries – 1) then
UNDEFINED
endif

TLBR
IRead Indexed TLB Entry
if ( (Config4IE ≥ 1) and TLB[i]VPN2_invalid = 1) then
PagemaskMask  0 // or value representing minimum page size EntryHi  0
EntryLo1  0
EntryLo0  0
EntryHiEHINV  1
else
PageMaskMask  TLB[i]Mask EntryHi 
(TLB[i]VPN2 and not TLB[i]Mask) || # Masking implem dependent
05 || TLB[i]ASID EntryLo1  02 ||
(TLB[i]PFN1 and not TLB[i]Mask) || # Masking mplem dependent
TLB[i]C1 || TLB[i]D1 || TLB[i]V1 || TLB[i]G EntryLo0  02 ||
endif
Exceptions:
(TLB[i]PFN0 and not TLB[i]Mask) || # Masking mplem dependent TLB[i]C0 || TLB[i]D0 || TLB[i]V0 || TLB[i]G
Coprocessor Unusable, Machine Check
The MIPS32® Instruction Set Manual, Revision 6.06 429

TLBWI Write Indexed TLB Entry
31
26 25 24 6 5 0
COP0 010000
CO 1
0
000 0000 0000 0000 0000
TLBWI 000010
430
The MIPS32® Instruction Set Manual, Revision 6.06
61 19 6
Format: TLBWI
Purpose: Write Indexed TLB Entry
To write or invalidate a TLB entry indexed by the Index register. Description:
If Config4IE == 0 or EntryHiEHINV=0:
MIPS32
The TLB entry pointed to by the Index register is written from the contents of the EntryHi, EntryLo0, EntryLo1, and PageMask registers. It is implementation dependent whether multiple TLB matches are detected on a TLBWI. In such an instance, a Machine Check Exception is signaled.
In Release 2 of the Architecture, multiple TLB matches may only be reported on a TLB write. The information written to the TLB entry may be different from that in the EntryHi, EntryLo0, and EntryLo1 registers, in that:
• The value written to the VPN2 field of the TLB entry may have those bits set to zero corresponding to the one bits in the Mask field of the PageMask register (the least significant bit of VPN2 corresponds to the least significant bit of the Mask field). It is implementation dependent whether these bits are preserved or zeroed during a TLB write.
• The value written to the PFN0 and PFN1 fields of the TLB entry may have those bits set to zero correspond- ing to the one bits in the Mask field of PageMask register (the least significant bit of PFN corresponds to the least significant bit of the Mask field). It is implementation dependent whether these bits are preserved or zeroed during a TLB write.
• The single G bit in the TLB entry is set from the logical AND of the G bits in the EntryLo0 and EntryLo1 registers.
If Config4IE ≥ 1 and EntryHiEHINV = 1:
The TLB entry pointed to by the Index register has its VPN2 field marked as invalid. This causes the entry to be
ignored on TLB matches for memory accesses. No Machine Check is generated.
Restrictions:
The operation is UNDEFINED if the contents of the Index register are greater than or equal to the number of TLB
entries in the processor.
If access to Coprocessor 0 is not enabled, a Coprocessor Unusable Exception is signaled.
Release 6: Processors that include a Block Address Translation (BAT) or Fixed Mapping (FM) MMU (ConfigMT = 2 or 3), the operation of this instruction causes a Reserved Instruction exception (RI).
Operation:
i  Index
if (Config4IE ≥ 1) then
TLB[i]VPN2_invalid  0
if ( EntryHIEHINV=1 ) then

TLBWI
IWrite Indexed TLB Entry
TLB[i]VPN2_invalid  1 break
endif endif
TLB[i]Mask 
TLB[i]VPN2 
TLB[i]ASID 
TLB[i]G  EntryLo1G and EntryLo0G
TLB[i]PFN1  EntryLo1PFN and not PageMaskMask # Implementation dependent TLB[i]C1  EntryLo1C
PageMaskMask
EntryHiVPN2 and not PageMaskMask # Implementation dependent EntryHiASID
TLB[i]D1  EntryLo1D
TLB[i]V1  EntryLo1V
TLB[i]PFN0  EntryLo0PFN and not PageMaskMask # Implementation dependent TLB[i]C0  EntryLo0C
TLB[i]D0  EntryLo0D
TLB[i]V0  EntryLo0V
Exceptions:
Coprocessor Unusable, Machine Check
The MIPS32® Instruction Set Manual, Revision 6.06 431

TLBWR Write Random TLB Entry
31
26 25 24 6 5 0
61 19 6
Format: TLBWR MIPS32 Purpose: Write Random TLB Entry
To write a TLB entry indexed by the Random register, or, in Release 6, write a TLB entry indexed by an implemen- tation-defined location.
Description:
The TLB entry pointed to by the Random register is written from the contents of the EntryHi, EntryLo0, EntryLo1, and PageMask registers. It is implementation dependent whether multiple TLB matches are detected on a TLBWR. In such an instance, a Machine Check Exception is signaled.
In Release 6, the Random register has been removed. References to Random refer to an implementation-determined value that is not visible to software.
In Release 2 of the Architecture, multiple TLB matches may only be reported on a TLB write. The information writ- ten to the TLB entry may be different from that in the EntryHi, EntryLo0, and EntryLo1 registers, in that:
• The value written to the VPN2 field of the TLB entry may have those bits set to zero corresponding to the one bits in the Mask field of the PageMask register (the least significant bit of VPN2 corresponds to the least signif- icant bit of the Mask field). It is implementation dependent whether these bits are preserved or zeroed during a TLB write.
• The value written to the PFN0 and PFN1 fields of the TLB entry may have those bits set to zero corresponding to the one bits in the Mask field of PageMask register (the least significant bit of PFN corresponds to the least sig- nificant bit of the Mask field). It is implementation dependent whether these bits are preserved or zeroed during a TLB write.
• The single G bit in the TLB entry is set from the logical AND of the G bits in the EntryLo0 and EntryLo1 regis- ters.
Restrictions:
If access to Coprocessor 0 is not enabled, a Coprocessor Unusable Exception is signaled.
Release 6: Processors that include a Block Address Translation (BAT) or Fixed Mapping (FM) MMU (ConfigMT = 2 or 3), the operation of this instruction causes a Reserved Instruction exception (RI).
Operation:
i  Random
if (Config4IE ≥ 1) then
TLB[i]VPN2_invalid  0 endif
TLB[i]Mask  PageMaskMask
TLB[i]VPN2  EntryHiVPN2 and not PageMaskMask # Implementation dependent TLB[i]ASID  EntryHiASID
TLB[i]G  EntryLo1G and EntryLo0G
TLB[i]PFN1  EntryLo1PFN and not PageMaskMask # Implementation dependent TLB[i]C1  EntryLo1C
TLB[i]D1  EntryLo1D
TLB[i]V1  EntryLo1V
TLB[i]PFN0  EntryLo0PFN and not PageMaskMask # Implementation dependent
COP0 010000
CO 1
0
000 0000 0000 0000 0000
TLBWR 000110
432
The MIPS32® Instruction Set Manual, Revision 6.06

TLBWR
IWrite Random TLB Entry
TLB[i]C0  EntryLo0C TLB[i]D0  EntryLo0D TLB[i]V0  EntryLo0V
Exceptions:
Coprocessor Unusable, Machine Check
The MIPS32® Instruction Set Manual, Revision 6.06
433

TLT
Trap if Less Than
31
26 25 21 20 16 15 6 5 0
6 5 5 10 6
Format: TLT rs, rt MIPS32 Purpose: Trap if Less Than
To compare GPRs and do a conditional trap.
Description: if GPR[rs] < GPR[rt] then Trap Compare the contents of GPR rs and GPR rt as signed integers. If GPR rs is less than GPR rt, then take a Trap excep- tion. The contents of the code field are ignored by hardware and may be used to encode information for system software. To retrieve the information, system software must load the instruction word from memory. Alternatively, if CP0 BadInstr is implemented, the code field may be obtained from BadInstr. Restrictions: None Operation: if GPR[rs] < GPR[rt] then SignalException(Trap) endif Exceptions: Trap SPECIAL 000000 rs rt code TLT 110010 434 The MIPS32® Instruction Set Manual, Revision 6.06 TLTI 31 ITrap if Less Than Immediate 0 26 25 21 20 16 15 655 16 Format: TLTI rs, immediate Purpose: Trap if Less Than Immediate To compare a GPR to a constant and do a conditional trap. MIPS32, removed in Release 6 REGIMM 000001 rs TLTI 01010 immediate Description: if GPR[rs]  immediate then Trap Compare the contents of GPR rs and the 16-bit signed immediate as signed integers. If GPR rs is less than immediate, then take a Trap exception. Restrictions: None Availability and Compatibility: This instruction has been removed in Release 6. Operation: if GPR[rs]  sign_extend(immediate) then SignalException(Trap) endif Exceptions: Trap The MIPS32® Instruction Set Manual, Revision 6.06 435 TLTIU Trap if Less Than Immediate Unsigned 31 26 25 21 20 16 15 655 16 Format: TLTIU rs, immediate Purpose: Trap if Less Than Immediate Unsigned To compare a GPR to a constant and do a conditional trap. 0 MIPS32, removed in Release 6 REGIMM 000001 rs TLTIU 01011 immediate 436 The MIPS32® Instruction Set Manual, Revision 6.06 Description: if GPR[rs]  immediate then Trap Compare the contents of GPR rs and the 16-bit sign-extended immediate as unsigned integers. If GPR rs is less than immediate, then take a Trap exception. Because the 16-bit immediate is sign-extended before comparison, the instruction can represent the smallest or largest unsigned numbers. The representable values are at the minimum [0, 32767] or maximum [max_unsigned-32767, max_unsigned] end of the unsigned range. Restrictions: None Availability and Compatibility: This instruction has been removed in Release 6. Operation: if (0 || GPR[rs])  (0 || sign_extend(immediate)) then SignalException(Trap) endif Exceptions: Trap TLTU ITrap if Less Than Unsigned 31 26 25 21 20 16 15 6 5 0 6 5 5 10 6 Format: TLTU rs, rt MIPS32 Purpose: Trap if Less Than Unsigned To compare GPRs and do a conditional trap. Description: if GPR[rs]  GPR[rt] then Trap Compare the contents of GPR rs and GPR rt as unsigned integers. If GPR rs is less than GPR rt, then take a Trap exception. The contents of the code field are ignored by hardware and may be used to encode information for system software. To retrieve the information, system software must load the instruction word from memory. Alternatively, if CP0 BadInstr is implemented, the code field may be obtained from BadInstr. Restrictions: None Operation: if (0 || GPR[rs])  (0 || GPR[rt]) then SignalException(Trap) endif Exceptions: Trap SPECIAL 000000 rs rt code TLTU 110011 The MIPS32® Instruction Set Manual, Revision 6.06 437 TNE Trap if Not Equal 31 26 25 21 20 16 15 6 5 0 6 5 5 10 6 Format: TNE rs, rt MIPS32 Purpose: Trap if Not Equal To compare GPRs and do a conditional trap. Description: if GPR[rs] ≠ GPR[rt] then Trap Compare the contents of GPR rs and GPR rt as signed integers. If GPR rs is not equal to GPR rt, then take a Trap exception. The contents of the code field are ignored by hardware and may be used to encode information for system software. To retrieve the information, system software must load the instruction word from memory. Alternatively, if CP0 BadInstr is implemented, the code field may be obtained from BadInstr. Restrictions: None Operation: if GPR[rs] ≠ GPR[rt] then SignalException(Trap) endif Exceptions: Trap SPECIAL 000000 rs rt code TNE 110110 438 The MIPS32® Instruction Set Manual, Revision 6.06 TNEI 31 ITrap if Not Equal Immediate 0 26 25 21 20 16 15 655 16 Format: TNEI rs, immediate Purpose: Trap if Not Equal Immediate To compare a GPR to a constant and do a conditional trap. MIPS32, removed in Release 6 REGIMM 000001 rs TNEI 01110 immediate Description: if GPR[rs]  immediate then Trap Compare the contents of GPR rs and the 16-bit signed immediate as signed integers. If GPR rs is not equal to imme- diate, then take a Trap exception. Restrictions: None Availability and Compatibility: This instruction has been removed in Release 6. Operation: if GPR[rs]  sign_extend(immediate) then SignalException(Trap) endif Exceptions: Trap The MIPS32® Instruction Set Manual, Revision 6.06 439 TRUNC.L.fmt Floating Point Truncate to Long Fixed Point 31 26 25 21 20 16 15 11 10 6 5 0 655556 COP1 010001 fmt 0 00000 fs fd TRUNC.L 001001 440 The MIPS32® Instruction Set Manual, Revision 6.06 Format: TRUNC.L.fmt TRUNC.L.S fd, fs TRUNC.L.D fd, fs Purpose: Floating Point Truncate to Long Fixed Point To convert an FP value to 64-bit fixed point, rounding toward zero. MIPS64,MIPS32 Release 2 MIPS64,MIPS32 Release 2 Description: FPR[fd]  convert_and_round(FPR[fs]) The value in FPR fs, in format fmt, is converted to a value in 64-bit long-fixed point format and rounded toward zero (rounding mode 1). The result is placed in FPR fd. When the source value is Infinity, NaN, or rounds to an integer outside the range -263 to 263-1, the result cannot be represented correctly and an IEEE Invalid Operation condition exists. In this case the Invalid Operation flag is set in the FCSR. If the Invalid Operation Enable bit is set in the FCSR, no result is written to fd and an Invalid Operation exception is taken immediately. Otherwise, a default result is written to fd. On cores with FCSRNAN2008=0, the default result is 263–1. On cores with FCSRNAN2008=1, the default result is: • 0 when the input value is NaN • 263–1 when the input value is + or rounds to a number larger than 263–1 • -263–1 when the input value is – or rounds to a number smaller than -263–1 Restrictions: The fields fs and fd must specify valid FPRs: fs for type fmt and fd for long fixed point. If the fields are not valid, the result is UNPREDICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. The result of this instruction is UNPREDICTABLE if the processor is executing in the FR=0 32-bit FPU register model; it is predictable if executing on a 64-bit FPU in the FR=1 mode, but not with FR=0, and not on a 32-bit FPU. Operation: StoreFPR(fd, L, ConvertFmt(ValueFPR(fs, fmt), fmt, L)) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Unimplemented Operation, Invalid Operation, Inexact TRUNC.W.fmt IFloating Point Truncate to Word Fixed Point 31 26 25 21 20 16 15 11 10 6 5 655556 Format: TRUNC.W.fmt TRUNC.W.S fd, fs TRUNC.W.D fd, fs Purpose: Floating Point Truncate to Word Fixed Point To convert an FP value to 32-bit fixed point, rounding toward zero. 0 MIPS32 MIPS32 COP1 010001 fmt 0 00000 fs fd TRUNC.W 001101 Description: FPR[fd]  convert_and_round(FPR[fs]) The value in FPR fs, in format fmt, is converted to a value in 32-bit word fixed point format using rounding toward zero (rounding mode 1). The result is placed in FPR fd. When the source value is Infinity, NaN, or rounds to an integer outside the range -231 to 231-1, the result cannot be represented correctly and an IEEE Invalid Operation condition exists. In this case the Invalid Operation flag is set in the FCSR. If the Invalid Operation Enable bit is set in the FCSR, no result is written to fd and an Invalid Operation exception is taken immediately. Otherwise, a default result is written to fd. On cores with FCSRNAN2008=0, the default result is 231–1. On cores with FCSRNAN2008=1, the default result is: • 0 when the input value is NaN • 231–1 when the input value is + or rounds to a number larger than 231–1 • -231–1 when the input value is – or rounds to a number smaller than -231–1 Restrictions: The fields fs and fd must specify valid FPRs: fs for type fmt and fd for word fixed point. If the fields are not valid, the result is UNPREDICTABLE. The operand must be a value in format fmt; if it is not, the result is UNPREDICTABLE and the value of the operand FPR becomes UNPREDICTABLE. Operation: StoreFPR(fd, W, ConvertFmt(ValueFPR(fs, fmt), fmt, W)) Exceptions: Coprocessor Unusable, Reserved Instruction Floating Point Exceptions: Inexact, Invalid Operation, Unimplemented Operation The MIPS32® Instruction Set Manual, Revision 6.06 441 WAIT Enter Standby Mode 31 26 25 24 6 5 0 COP0 010000 CO 1 Implementation-dependent code WAIT 100000 442 The MIPS32® Instruction Set Manual, Revision 6.06 61 19 6 Format: WAIT Purpose: Enter Standby Mode Wait for Event Description: MIPS32 The WAIT instruction performs an implementation-dependent operation, involving a lower power mode. Software may use the code bits of the instruction to communicate additional information to the processor. The processor may use this information as control for the lower power mode. A value of zero for code bits is the default and must be valid in all implementations. The WAIT instruction is implemented by stalling the pipeline at the completion of the instruction and entering a lower power mode. The pipeline is restarted when an external event, such as an interrupt or external request occurs, and execution continues with the instruction following the WAIT instruction. It is implementation-dependent whether the pipeline restarts when a non-enabled interrupt is requested. In this case, software must poll for the cause of the restart. The assertion of any reset or NMI must restart the pipeline and the corresponding exception must be taken. If the pipeline restarts as the result of an enabled interrupt, that interrupt is taken between the WAIT instruction and the following instruction (EPC for the interrupt points at the instruction following the WAIT instruction). In Release 6, the behavior of WAIT has been modified to make it a requirement that a processor that has disabled operation as a result of executing a WAIT will resume operation on arrival of an interrupt even if interrupts are not enabled. In Release 6, the encoding of WAIT with bits 24:6 of the opcode set to 0 will never disable CP0 Count on an active WAIT instruction. In particular, this modification has been added to architecturally specify that CP0 Count is not dis- abled on execution of WAIT with default code of 0. Prior to Release 6, whether Count is disabled was implementa- tion-dependent. In the future, other encodings of WAIT may be defined which specify other forms of power-saving or stand-by modes. If not implemented, then such unimplemented encodings must default to WAIT 0. Restrictions: Pre-Release 6: The operation of the processor is UNDEFINED if a WAIT instruction is executed in the delay slot of a branch or jump instruction. Release 6: Implementations are required to signal a Reserved Instruction exception if WAIT is encountered in the delay slot or forbidden slot of a branch or jump instruction. If access to Coprocessor 0 is not enabled, a Coprocessor Unusable Exception is signaled. Operation: Pre-Release 6: I: Enter implementation dependent lower power mode I+1:/* Potential interrupt taken here */ Release 6: I: if IsCoprocessorEnabled(0) then while ( !interrupt_pending_and_not_masked_out() && !implementation_dependent_wake_event() ) < enter or remain in low power mode or stand-by mode>

WAIT
IEnter Standby Mode
else
SignalException(CoprocessorUnusable, 0)
endif
I+1: if ( interrupt_pending() && interrupts_enabled() ) then EPC  PC + 4
< process interrupt; execute ERET eventually >
else
endif
// unblock on non-enabled interrupt or imp dep wake event. PC  PC + 4
< continue execution at instruction after wait >
function interrupt_pending_and_not_masked_out
return (Config3VEIC && IntCtlVS && CauseIV && !StatusBEV)
? CauseRIPL > StatusIPL : CauseIP & StatusIM;
endfunction
function interrupts_enabled
return StatusIE && !StatusEXL && !StatusERL && !DebugDM;
endfunction
function implementation_dependent_wake_event

endfunction
Exceptions:
Coprocessor Unusable Exception
The MIPS32® Instruction Set Manual, Revision 6.06 443

WRPGPR Write to GPR in Previous Shadow Set
31
26 25 21 20 16 15 11 10
6 5 5 5 11 Format: WRPGPR rd, rt
Purpose: Write to GPR in Previous Shadow Set
To move the contents of a current GPR to a GPR in the previous shadow set.
0
MIPS32 Release 2
COP0 0100 00
WRPGPR 01 110
rt
rd
0
000 0000 0000
444
The MIPS32® Instruction Set Manual, Revision 6.06
Description: SGPR[SRSCtlPSS, rd]  GPR[rt]
The contents of the current GPR rt is moved to the shadow GPR register specified by SRSCtlPSS (signifying the pre-
vious shadow set number) and rd (specifying the register number within that set). Restrictions:
In implementations prior to Release 2 of the Architecture, this instruction resulted in a Reserved Instruction excep- tion.
Operation:
SGPR[SRSCtlPSS, rd]  GPR[rt]
Exceptions:
Coprocessor Unusable, Reserved Instruction

WSBH IWord Swap Bytes Within Halfwords
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: WSBH rd, rt MIPS32 Release 2 Purpose: Word Swap Bytes Within Halfwords
To swap the bytes within each halfword of GPR rt and store the value into GPR rd. Description: GPR[rd]  SwapBytesWithinHalfwords(GPR[rt])
Within each halfword of GPR rt the bytes are swapped, and stored in GPR rd. Restrictions:
In implementations prior to Release 2 of the architecture, this instruction resulted in a Reserved Instruction exception.
sOperation:
GPR[rd]  GPR[rs]23..16 || GPR[rs]31..24 || GPR[rs]7..0 || GPR[rs]15..8
Exceptions:
Reserved Instruction
Programming Notes:
The WSBH instruction can be used to convert halfword and word data of one endianness to another endianness. The endianness of a word value can be converted using the following sequence:
SPECIAL3 011111
0 00000
rt
rd
WSBH 00010
BSHFL 100000
lw t0, 0(a1) wsbh t0, t0 rotr t0, t0, 16
/* Read word value */
/* Convert endiannes of the halfwords */
/* Swap the halfwords within the words */
Combined with SEH and SRA, two contiguous halfwords can be loaded from memory, have their endianness con- verted, and be sign-extended into two word values in four instructions. For example:
lw t0, 0(a1) wsbh t0, t0 seh t1, t0 sra t0, t0, 16
/* Read two contiguous halfwords */
/* Convert endiannes of the halfwords */
/* t1 = lower halfword sign-extended to word */
/* t0 = upper halfword sign-extended to word */
Zero-extended words can be created by changing the SEH and SRA instructions to ANDI and SRL instructions, respectively.
.
The MIPS32® Instruction Set Manual, Revision 6.06 445

XOR
Exclusive OR
31
26 25 21 20 16 15 11 10 6 5 0
655556
Format: XOR rd, rs, rt MIPS32 Purpose: Exclusive OR
To do a bitwise logical Exclusive OR.
Description: GPR[rd]  GPR[rs] XOR GPR[rt]
Combine the contents of GPR rs and GPR rt in a bitwise logical Exclusive OR operation and place the result into
GPR rd. Restrictions:
None
Operation:
GPR[rd]  GPR[rs] xor GPR[rt]
Exceptions:
None
SPECIAL 000000
rs
rt
rd
0 00000
XOR 100110
446
The MIPS32® Instruction Set Manual, Revision 6.06

XORI
31
26 25 21 20 16 15
655 16
Format: XORI rt, rs, immediate Purpose: Exclusive OR Immediate
To do a bitwise logical Exclusive OR with a constant.
IExclusive OR Immediate 0
MIPS32
XORI 001110
rs
rt
immediate
Description: GPR[rt]  GPR[rs] XOR immediate
Combine the contents of GPR rs and the 16-bit zero-extended immediate in a bitwise logical Exclusive OR operation
and place the result into GPR rt. Restrictions:
None
Operation:
GPR[rt]  GPR[rs] xor zero_extend(immediate) Exceptions:
None
The MIPS32® Instruction Set Manual, Revision 6.06 447

Appendix A
Instruction Bit Encodings
A.1 Instruction Encodings and Instruction Classes
Instruction encodings are presented in this section; field names are printed here and throughout the book in italics. When encoding an instruction, the primary opcode field is encoded first. Most opcode values completely specify an
instruction that has an immediate value or offset.
Opcode values that do not specify an instruction instead specify an instruction class. Instructions within a class are further specified by values in other fields. For instance, opcode REGIMM specifies the immediate instruction class, which includes conditional branch and trap immediate instructions.
A.2 Instruction Bit Encoding Tables
This section provides various bit encoding tables for the instructions of the MIPS32® ISA.
Figure A.1 shows a sample encoding table and the instruction opcode field this table encodes. Bits 31..29 of the opcode field are listed in the leftmost columns of the table. Bits 28..26 of the opcode field are listed along the topmost rows of the table. Both decimal and binary values are given, with the first three bits designating the row, and the last three bits designating the column.
An instruction’s encoding is found at the intersection of a row (bits 31..29) and column (bits 28..26) value. For instance, the opcode value for the instruction labeled EX1 is 33 (decimal, row and column), or 011011 (binary). Sim- ilarly, the opcode value for EX2 is 64 (decimal), or 110100 (binary).
Release 6 introduces additional nomenclature to the opcode tables for Release 6 instructions. For new instructions, bits 31:26 are generically named POPXY where X is the row number, and Y is the column number. This convention is extended to sub-opcode tables, except bits 5:0 are generically named SOPXY, where X is the row number, and Y is the column number. This naming convention is applied where a specific encoded value may be shared by multiple instructions.
The MIPS32® Instruction Set Manual, Revision 6.06 448

Instruction Bit Encodings
Table A.1 Symbols Used in the Instruction Encoding Tables (Continued)
Symbol
Meaning

Operation or field codes marked with this symbol represent instructions which were only legal if 64-bit operations were enabled on implementations of Release 1 of the Architecture. In Release 2 of the architecture, operation or field codes marked with this symbol represent instructions which are legal if 64-bit floating point operations are enabled. In other cases, executing such an instruction must cause a Reserved Instruction exception (non-coprocessor encodings or coprocessor instruction encod- ings for a coprocessor to which access is allowed) or a Coprocessor Unusable Excep- tion (coprocessor instruction encodings for a coprocessor to which access is not allowed).

Instructions formerly marked  in some earlier versions of manuals, corrected and marked  in revision 5.03. Legal on MIPS64r1 but not MIPS32r1; in release 2 and above, legal in both MIPS64 and MIPS32, in particular even when running in “32-bit FPU Register File mode”, FR=0, as well as FR=1.

Operation or field codes marked with this symbol are available to licensed MIPS part- ners. To avoid multiple conflicting instruction definitions, MIPS Technologies will assist the partner in selecting appropriate encodings if requested by the partner. The partner is not required to consult with MIPS Technologies when one of these encod- ings is used. If no instruction is encoded with this value, executing such an instruction must cause a Reserved Instruction exception (SPECIAL2 encodings or coprocessor instruction encodings for a coprocessor to which access is allowed) or a Coprocessor Unusable Exception (coprocessor instruction encodings for a coprocessor to which access is not allowed).

Release 6 reserves the SPECIAL2 encodings. pre-MIPS32 Release 2 the SPECIAL2 encodings were available for customer use as UDIs. Otherwise like  above.

Field codes marked with this symbol represent an EJTAG support instruction and implementation of this encoding is optional for each implementation. If the encoding is not implemented, executing such an instruction must cause a Reserved Instruction exception. If the encoding is implemented, it must match the instruction encoding as shown in the table.

Operation or field codes marked with this symbol are reserved for MIPS optional Module or Application Specific Extensions. If the Module/ASE is not implemented, executing such an instruction must cause a Reserved Instruction exception.

Operation or field codes marked with this symbol are obsolete and will be removed from a future revision of the MIPS32 ISA. Software should avoid using these opera- tion or field codes.

Operation or field codes marked with this symbol are valid for Release 2 implementa- tions of the architecture. Executing such an instruction in a Release 1 implementation must cause a Reserved Instruction exception.
6N
Instruction added by Release 6. “N” for “new”.
6Nm
New Release 6 encoding for a pre-Release 6 instruction that has been moved. “Nm” for “New (moved)
450 The MIPS32® Instruction Set Manual, Revision 6.06

Table A.1 Symbols Used in the Instruction Encoding Tables (Continued)
A.2 Instruction Bit Encoding Tables
Symbol
Meaning
6Rm
pre-Release 6 instruction encoding moved in Release 6. “Rm” for “Removed (moved elsewhere)”.
6Rm and 6R instructions signal a Reserved Instruc- tion exception when exe- cuted by a Release 6 implementation. If the encoding has been used for a new instruction or coprocessor, the unus- able exception takes pri- ority.
6R
pre-Release 6 instruction encoding removed by Release 6. “R” for “Removed”.
Table A.2 MIPS32 Encoding of the Opcode Field
bits 28..26 bits 31..29
opcode
0
1
2
3
4
5
6
7
000
001
010
011
100
101
110
111
0
000
SPECIAL 
ADDI6R POP106N 
COP0 
POP306N
LB
SB
LL6Rm
SC6Rm
1
001
REGIMM 
ADDIU
COP1 

LH
SH
LWC1
SWC1
2
010
J
SLTI
COP2 

LWL6R
SWL6R
LWC26Rm  BC6N
SWC26Rm 
BALC6N 
3
011
JAL
SLTIU
COP1X2 6R

LW
SW
PREF6Rm
PCREL6N
4
100
BEQ
ANDI
BEQL6R 
SPECIAL2 6R  
LBU



5
101
BNE
ORI
BNEL6R 
JALX6R 
LHU

LDC1
SDC1
6
110
BLEZ POP066N 
XORI
BLEZL6R  POP266N 
MSA 
LWR6R
SWR6R
LDC26Rm  BEQZC/JIC6N POP666N 
SDC26Rm  BNEZC/JIALC6N POP766N
7
111
BGTZ POP076N 
LUI1 / AUI6N
BGTZL6R  POP276N 
SPECIAL33 

CACHE6Rm


1. Pre-Release 6 instruction LUI is a special case of Release 6 instruction AUI.
2. Architecture Release 1, the COP1X opcode was called COP3, and was available as another user-available coproces-
sor. Architecture Release 2, a full 64-bit floating point unit is available with 32-bit CPUs, and the COP1X opcode is reserved for that purpose on all Release 2 CPUs. 32-bit implementations of Release 1 of the architecture are strongly discouraged from using this opcode for a user-available coprocessor as doing so limits the potential for an upgrade path for the FPU.
3. Architecture Release 2 added the SPECIAL3 opcode. Implementations of Release 1 of the Architecture signaled a Reserved Instruction exception for this opcode.
The MIPS32® Instruction Set Manual, Revision 6.06 451

Instruction Bit Encodings
bits 2..0 bits 5. 3
function
Table A.3 MIPS32 SPECIAL Opcode Encoding of Function Field
0
1
2
3
4
5
6
7
000
001
010
0114
100
101
110
111
0
000
SLL1
JR2,3,6R
MFHI6R CLZ6Nm
4MULT6R SOP306N
ADD

TGE

1
001
MOVCI 6R
JALR2
MTHI6R CLO6Nm
4MULTU6R SOP316N
ADDU

TGEU

2
010
SRL 
MOVZ6R
MFLO6R
4DIV6R SOP326N
SUB
SLT
TLT

3
011
SRA
MOVN6R
MTLO6R
4DIVU6R SOP336N
SUBU
SLTU
TLTU

4
100
SLLV
SYSCALL


AND

TEQ

5
101
LSA6N
BREAK


OR

SELEQZ6N

6
110
SRLV 
SDBBP6Nm


XOR

TNE

7
111
SRAV
SYNC


NOR

SELNEZ6N

1. Specific encodings of the rt, rd, and sa fields are used to distinguish among the SLL, NOP, SSNOP, EHB and PAUSE functions. Release 6 makes SSNOP equivalent to NOP.
2. Specific encodings of the hint field are used to distinguish JR from JR.HB and JALR from JALR.HB
3. Release 6 removes JR and JR.HB. JALR with rd=0 provides functionality equivalent to JR. JALR.HB with rd=0 provides func-
tionality equivalent to JR.HB. Assemblers should produce the new instruction when encountering the old mnemonic.
4. Specific encodings of the sa field are used to distinguish pre-Release 6 and Release 6 integer multiply and divide instructions.
See Table A.23 on page 462, which shows that the encodings do not conflict. The pre-Release 6 divide instructions signal Reserved Instruction exception on Release 6. Note that the same mnemonics are used for pre-Release 6 divide instructions that return both quotient and remainder, and Release 6 divide instructions that return only quotient, with separate MOD instructions for the remainder.
bits 18..16 bits 20..19
1. NAL and BAL are assembly idioms prior to Release 6.
rt
Table A.4 MIPS32 REGIMM Encoding of rt Field
0
1
2
3
00
01
10
11
0
000
BLTZ
TGEI6R
BLTZAL6R NAL6N 1

1
001
BGEZ
TGEIU6R
BGEZAL6R BAL6N 1

2
010
BLTZL6R 
TLTI6R
BLTZALL6R 

3
011
BGEZL6R 
TLTIU6R
BGEZALL6R 

4
100

TEQI6R


5
101




6
110
DAHI6N
TNEI6R

DATI6N
7
111


SIGRIE6N
SYNCI 
452
The MIPS32® Instruction Set Manual, Revision 6.06

Table A.5 MIPS32 SPECIAL2 Encoding of Function Field bits 2..0
A.2 Instruction Bit Encoding Tables
function
bits 5..3
0
1
2
3
4
5
6
7
000
001
010
011
100
101
110
111
0
000
MADD6R

 


CLZ6Rm



1
001
MADDU6R




CLO6Rm



2
010
MUL6R








3
011








4
100
MSUB6R








5
101
MSUBU6R








6
110








7
111







SDBBP6Rm
 
Table A.6 MIPS32 SPECIAL31 Encoding of Function Field for Release 2 of the Architecture bits 2..0
bits 5..3
function
0
1
2
3
4
5
6
7
000
001
010
011
100
101
110
111
0
000
EXT 



BSHFL 
LBUE


1
001



LWLE6R
SWLE6R
LHUE


2
010



LWRE6R
SWRE6R



3
011



CACHEE
PREFE
*

RDHWR 
4
100
INS 


SBE

LBE


5
101



SHE
CACHE6Nm
LHE
PREF6Nm

6
110



SCE
SC6Nm
LLE
LL6Nm

7
111



SWE

LWE


1.Architecture Release 2 added the SPECIAL3 opcode. Implementations of Release 1 of the Architecture signaled a Reserved Instruction exception for this opcode and all function field values shown above.
Table A.7 MIPS32 MOVCI6R1 Encoding of tf Bit bit 16
1. Release 6 removes the MOVCI instruction family (MOVT and MOVF).
tf
0
1
MOVF6R
MOVT6R
The MIPS32® Instruction Set Manual, Revision 6.06 453

Instruction Bit Encodings
Table A.8 MIPS321 SRL Encoding of Shift/Rotate bit 21
1. Release 2 of the Architecture added the ROTR instruction. Implementations of Release 1 of the Architecture ignored bit 21 and treated the instruction as an SRL
Table A.9 MIPS321 SRLV Encoding of Shift/Rotate bit 6
1. Release 2 of the Architecture added the ROTRV instruction. Implementations of Release 1 of the Architecture ignored bit 6 and treated the instruction as an SRLV
Table A.10 MIPS32 BSHFL Encoding of sa Field1
R
0
1
SRL
ROTR
R
0
1
SRLV
ROTRV
bits 8..6 bits 10..9
sa
0
1
2
3
00
01
10
11
0
000
BITSWAP6N 6N
SEB
SEH
1
001
*
ALIGN6N (BSHFL)
*
*
2
010
WSBH
*
*
3
011
*
*
*
4
100
*
*
*
*
5
101
*
*
*
6
110
*
*
*
*
7
111
*
*
*
*
454
The MIPS32® Instruction Set Manual, Revision 6.06
1. The sa field is sparsely decoded to identify the final instructions. Entries in this table with no mnemonic are reserved for future use by MIPS technologies and may or may not cause a Reserved Instruction exception.

bits 23..21 bits 25..24
rs
Table A.11 MIPS32 COP0 Encoding of rs Field
A.2 Instruction Bit Encoding Tables
0
1
2
3
00
01
10
11
0
000
MFC0

1
001


2
010
MFH
RDPGPR 
3
011

MFMC01 
C0 
4
100
MTC0

5
101


6
110
MTH
WRPGPR 
7
111


1. Release 2 of the Architecture added the MFMC0 function, which is further decoded as the DI (bit 5 = 0) and EI (bit 5 = 1) instructions.
Table A.12 MIPS32 COP0 Encoding of Function Field When rs=CO bits 2..0
function
bits 5..3
0
1
2
3
4
5
6
7
000
001
010
011
100
101
110
111
0
000

TLBP

ERET
WAIT



1
001
TLBR







2
010
TLBWI







3
011
TLBINV







4
100
TLBINVF







5
101








6
110
TLBWR







7
111



DERET 




Extension
bit 20..18 bit 17..16
Table A.13 PCREL Encoding of Minor Opcode Field
0
1
2
3
00
01
10
11
0
000
ADDIUPC
ADDIUPC
ADDIUPC
ADDIUPC
1
001
ADDIUPC
ADDIUPC
ADDIUPC
ADDIUPC
2
010
LWPC
LWPC
LWPC
LWPC
3
011
LWPC
LWPC
LWPC
LWPC
4
100
LWUPC
LWUPC
LWUPC
LWUPC
5
101
LWUPC
LWUPC
LWUPC
LWUPC
6
110
LDPC
LDPC
LDPC
7
111
*
*
AUIPC
LDPC
ALUIPC
The MIPS32® Instruction Set Manual, Revision 6.06 455

Instruction Bit Encodings
bits 23..21 bits 25..24
rs
Table A.14 MIPS32 Encoding of rs Field
0
1
2
3
00
01
10
11
0
000
MFC1
BC16R 
S
BZ.B 
1
001

BC1ANY26R  BC1EQZ6N
D
BZ.H 
2
010
CFC1
BC1ANY46R


BZ.W 
3
011
MFHC1 
BZ.V 

BZ.D 
4
100
MTC1

W
BNZ.B 
5
101

BC1NEZ6N
L
BNZ.H 
6
110
CTC1

PS6R  
BNZ.W 
7
111
MTHC1 
BNZ.V 

BNZ.D 
Table A.15 MIPS32 COP1 Encoding of Function Field When rs=S bits 2..0
function
bits 5..3
0
1
2
3
4
5
6
7
000
001
010
011
100
101
110
111
0
000
ADD
ROUND.L 
SEL 6N
MADDF 6N


C.F 6R CABS.F 
C.SF 6R CABS.SF 
1
001
SUB
TRUNC.L 
MOVCF 6R 
MSUBF 6N
CVT.D

C.UN 6R CABS.UN 
C.NGLE 6R CABS.NGLE 
2
010
MUL
CEIL.L 
MOVZ 6R
RINT 6N


C.EQ 6R CABS.EQ 
C.SEQ 6R CABS.SEQ 
3
011
DIV
FLOOR.L 
MOVN 6R
CLASS 6N


C.UEQ 6R CABS.UEQ 
C.NGL 6R CABS.NGL 
4
100
SQRT
ROUND.W
SELEQZ 6N
RECIP2 6R MIN 6N
CVT.W

C.OLT 6R CABS.OLT 
C.LT 6R CABS.LT 
5
101
ABS
TRUNC.W
RECIP 
RECIP1 6R MAX 6N
CVT.L 

C.ULT 6R CABS.ULT 
C.NGE 6R CABS.NGE 
6
110
MOV
CEIL.W
RSQRT 
RSQRT1 6R MINA 6N
CVT.PS 6R

C.OLE 6R CABS.OLE 
C.LE 6R CABS.LE 
7
111
NEG
FLOOR.W
SELNEZ 6N
RSQRT2  6R MAXA 6N


C.ULE 6R CABS.ULE 
C.NGT 6R CABS.NGT 
456
The MIPS32® Instruction Set Manual, Revision 6.06

Table A.16 MIPS32 COP1 Encoding of Function Field When rs=D bits 2..0
A.2 Instruction Bit Encoding Tables
function
bits 5..3
0
1
2
3
4
5
6
7
000
001
010
011
100
101
110
111
0
000
ADD
ROUND.L 
SEL6N
MADDF6N
CVT.S

C.F6R CABS.F 
C.SF6R CABS.SF 
1
001
SUB
TRUNC.L 
MOVCF6R 
MSUBF6N


C.UN6R CABS.UN 
C.NGLE6R CABS.NGLE 
2
010
MUL
CEIL.L 
MOVZ6R
RINT6N


C.EQ6R CABS.EQ 
C.SEQ6R CABS.SEQ 
3
011
DIV
FLOOR.L 
MOVN6R
CLASS6N


C.UEQ6R CABS.UEQ 
C.NGL6R CABS.NGL 
4
100
SQRT
ROUND.W
SELEQZ6N
RECIP2 6R MIN6N
CVT.W

C.OLT6R CABS.OLT 
C.LT6R CABS.LT 
5
101
ABS
TRUNC.W
RECIP 
RECIP1 6R MAX6N
CVT.L 
*
C.ULT6R CABS.ULT 
C.NGE6R CABS.NGE 
6
110
MOV
CEIL.W
RSQRT 
RSQRT1 6R MINA6N


C.OLE6R CABS.OLE 
C.LE6R CABS.LE 
7
111
NEG
FLOOR.W
SELNEZ6N
RSQRT2 6R MAXA6N


C.ULE6R CABS.ULE 
C.NGT6R CABS.NGT 
Table A.17 MIPS32 COP1 Encoding of Function Field When rs=W or L1 2 bits 2..0
function
bits 5..3
0
1
2
3
4
5
6
7
000
001
010
011
100
101
110
111
0
000
CMP.AF.S/D6N
CMP.SAF.S/D6N


CVT.S

1
001
CMP.UN.S/D6N
CMP.SUB.S/D6N
CMP.OR.S/D6N
CMP.SOR.S/D6N
CVT.D

2
010
CMP.EQ.S/D6N
CMP.SEQ.S/D6N
CMP.UNE.S/D6N
CMP.SUNE.S/D6N


3
011
CMP.UEQ.S/D6N
CMP.SUEQ.S/D6N
CMP.NE.S/D6N
CMP.SNE.S/D6N


4
100
CMP.OLT.S/D6N
CMP.SLT.S/D6N




5
101
CMP.ULT.S/D6N
CMP.SULT.S/D6N




6
110
CMP.OLE.S/D6N
CMP.SLE.S/D6N


CVT.PS.PW6R

7
111
CMP.ULE.S/D6N
CMP.SULE.S/D6N




1. Format type L is legal only if 64-bit floating point operations are enabled.
2. Release 6 introduces the CMP.condn fmt instruction family, where fmt=S or D, 32 or 64 bit floating point. However, .S and .D for
CMP.condn fmt are encoded as .W 10100 and .L 10101 in the “standard” format. The conditions tested are encoded the same way for pre-Release 6 C.cond.fmt and Release 6 CMP.cond fmt, except that Release 6 adds new conditions not present in C.cond fmt. Release 6, however, has changed the recommended mnemonics for the CMP.condn fmt to be consistent with the IEEE standard rather than pre-Release 6. See the table in the description of CMP.cond.fmt in Volume II of the MIPS Architecture Reference Man- ual, which shows the correspondence between pre-Release 6 C.cond fmt, Release 6 CMP.cond fmt, and MSA FC*.fmt / FS*.fmt floating point comparisons.
The MIPS32® Instruction Set Manual, Revision 6.06 457

Instruction Bit Encodings
Table A.18 MIPS32 COP1 Encoding of Function Field When rs=PS1 2 bits 2..0
function
bits 5..3
0
1
2
3
4
5
6
7
000
001
010
011
100
101
110
111
0
000
ADD6R


ADDR6R
CVT.S.PU6R
CVT.PS6R 
C.F.PS6R CABS.F.PS 
C.SF.PS6R CABS.SF.PS 
1
001
SUB6R

MOVCF6R



C.UN.PS6R CABS.UN 
C.NGLE.PS6R CABS.NGLE.PS 
2
010
MUL6R

MOVZ6R 
MULR6R


C.EQ6R CABS.EQ 
C.SEQ.PS6R CABS.SEQ.PS 
3
011


MOVN6R 



C.UEQ.PS6R CABS.UEQ.PS 
C.NGL.PS6R CABS.NGL.PS 
4
100



RECIP26R
CVT.PW.PS6R
PLL6R 
C.OLT.PS6R CABS.OLT.PS 
C.LT.PS6R CABS.LT.PS 
5
101
ABS6R


RECIP16R

PLU6R 
C.ULT6R CABS.ULT 
C.NGE.PS6R CABS.NGE.PS 
6
110
MOV6R 


RSQRT16R

PUL.PS6R 
C.OLE6R CABS.OLE 
C.LE.PS6R CABS.LE.PS 
7
111
NEG6R


RSQRT26R

PUU.PS6R 
C.ULE.PS6R CABS.ULE.PS 
C.NGT.PS6R CABS.NGT.PS 
1. Format type PS is legal only if 64-bit floating point operations are enabled. All encodings in this table are reserved in Release 6. 2. Release 6 removes format type PS (paired single). MSA (MIPS SIMD Architecture) may be used instead.
Table A.19 MIPS32 COP1 Encoding of tf Bit When rs=S, D, or PS6R, Function=MOVCF6R1 bit 16
0
1
MOVF.fmt6R
MOVT.fmt6R
bits 23..21 bits 25..24
tf
1. Release 6 removes the MOVCF instruction family (MOVF fmt and MOVT.fmt), replacing them by SEL fmt.
Table A.20 MIPS32 COP2 Encoding of rs Field
rs
0
1
2
3
00
01
10
11
0
000
MFC2 
BC26R 
1
001

BC2EQZ6N
2
010
CFC2 
LWC26Nm
3
011
MFHC2 
SWC26Nm
C2 
4
100
MTC2 

5
101

BC2NEZ6N
6
110
CTC2 
LDC26Nm
7
111
MTHC2 
SDC26Nm
458
The MIPS32® Instruction Set Manual, Revision 6.06

Table A.21 MIPS32 COP1X6R1 Encoding of Function Field bits 2..0
A.3 Floating Point Unit Instruction Format Encodings
function
bits 5..3
0
1
2
3
4
5
6
7
000
001
010
011
100
101
110
111
0
000
LWXC16R 
SWXC16R 


MADD.S6R2
MSUB.S6R
NMADD.S6R
NMSUB.S6R
1
001
LDXC16R 
SDXC16R 


MADD.D6R
MSUB.D6R
NMADD.D6R
NMSUB.D6R
2
010








3
011








4
100








5
101
LUXC16R 
SUXC16R 






6
110



ALNV.PS6R
MADD.PS6R
MSUB.PS6R
NMADD.PS6R
NMSUB.PS6R
7
111

PREFX6R 






1. Release 6 removes format type PS (paired single). MSA (MIPS SIMD Architecture) may be used instead.
2. Release 6 removes all pre-Release 6 COP1X instructions, of the form 010011 – COP1X.PS, non-fused FP multiply
adds, and indexed and unaligned loads, stores, and prefetches.
A.3 Floating Point Unit Instruction Format Encodings
Instruction format encodings for the floating point unit are presented in this section. This information is a tabular pre- sentation of the encodings described in tables ranging from Table A.14 to Table A.21 above.
Table A.22 Floating Point Unit Instruction Format Encodings
fmt field
(bits 25..21 of COP1 opcode)
fmt3 field (bits 2..0 of COP1X opcode)
Mnemonic
Name
Bit Width
Data Type
Decimal
Hex
Decimal
Hex
0..15
00..0F


Used to encode Coprocessor 1 interface instructions (MFC1, CTC1, etc.). Not used for format encoding.
16
10
0
0
S Single 32 Floating Point See note below: Release 6 CMP.condn.S/D encoded as W/L.
17
11
1
1
D Double 64 Floating Point See note below: Release 6 CMP.condn.S/D encoded as W/L.
18..19
12..13
2..3
2..3
Reserved for future use by the architecture.
20
14
4
4
W Word 32 Fixed Point See note below: Release 6 CMP.condn.S/D encoded as W/L.
21
15
5
5
L Long 64 Fixed Point See note below: Release 6 CMP.condn.S/D encoded as W/L.
22
16
6
6
PS Paired Single 2  32 Floating Point Release 6 removes the PS format, and reserves it for future use
23
17
7
7
Reserved for future use by the architecture.
The MIPS32® Instruction Set Manual, Revision 6.06 459

Instruction Bit Encodings
Table A.22 Floating Point Unit Instruction Format Encodings
Note: Release 6 CMP.condn.S/D encoded as W/L: as described in Table A.17 on page 457, “MIPS32 COP1 Encoding of Function Field When rs=W or L” on page 457, Release 6 uses certain instruction encodings with the rs (fmt) field equal to 11000 (W) or 11001 (L) to represent S and D respectively, for the instruction family CMP.condn fmt.
fmt field
(bits 25..21 of COP1 opcode)
fmt3 field (bits 2..0 of COP1X opcode)
Mnemonic
Name
Bit Width
Data Type
Decimal
Hex
Decimal
Hex
24..31
18..1F


Reserved for future use by the architecture. Not available for fmt3 encoding.
460
The MIPS32® Instruction Set Manual, Revision 6.06

A.4 Release 6 Instruction Encodings
Release 6 adds several new instructions, removes several old instructions, and changes the encodings of several pre- Release 6 instructions. In many cases, the old encodings for instructions moved or removed are required to signal the Reserved Instruction on Release 6, so that uses of old instructions can be trapped, and emulated or warned about; but in several cases the old encodings have been reused for new Release 6 instructions.
These instruction encoding changes are indicated in the tables above. Release 6 new instructions are superscripted 6N; Release 6 removed instructions are superscripted 6R; Release 6 instructions that have been moved are marked 6Rm at the pre-Release 6 encoding that they are moved from, and 6Nm at the new Release 6 encoding that it is moved to. Encoding table cells that contain both a non-Release 6 instruction and a Release 6 instruction superscripted 6N or 6Nm indicate a possible conflict, although in many cases footnotes indicate that other fields allow the distinction to be made.
The tables below show the further decoding in Release 6 for field classes (instruction encoding families) indicated in other tables.
Instruction encodings are also illustrated in the instruction descriptions in Volume II. Those encodings are authorita- tive. The instruction encoding tables in this section, above, based on bitfields, are illustrative, since they cannot com- pletely indicate the new tighter encodings.
MUL/DIV family encodings: Table A.23 below shows the Release 6 integer family of multiply and divide instruc- tions encodings, as well as the pre-Release 6 instructions they replace. The Release 6 and pre-Release 6 instructions share the same primary opcode, bits 31-26 = 000000, and share the function code, bits 5-0, with their pre-Release 6 counterparts, but are distinguished by bits 10-6 of the instruction. The pre-Release 6 instructions signal a Reserved Instruction exception on Release 6 implementations.
However, the instruction names collide: pre-Release 6 and Release 6 DIV, DIVU, DDIV, DDIVU are actually distinct instructions, although they share the same mnemonics. The pre-Release 6 instructions produce two results, both quo- tient and remainder in the HI/LO register pair, while the Release 6 DIV instruction produce only a single result, the quotient. It is possible to distinguish the conflicting instructions in assembly by looking at how many register oper- ands the instructions have, two versus three.
As of Release 6, all of pre-Release 6 instruction encodings that are removed are required to signal the reserved instruction exception, as are all in the vicinity 000000.xxxxx.xxxxx.aaaaa.011xxx, i.e. all with the primary opcodes and function codes listed in Table A.23, with the exception of the aaaaa field values 00010 and 00011 for the new instructions.
The MIPS32® Instruction Set Manual, Revision 6.06 461
A.4 Release 6 Instruction Encodings

Instruction Bit Encodings
Table A.23 Release 6 MUL/DIV encodings
pre-Release 6 removed struck through
00000.rs.rt.rd.aaaaa.function6
function bits 5-0
aaaaa, bits 10-6
00000
and rd = 00000 00010 00011
(bits 15-11)
011 000
MULT6R
MUL6N
MUH6N
011 001
MULTU6R
MULU6N
MUHU6N
011 010
DIV6R
DIV6N
MOD6N
011 011
DIVU6R
DIVU6N
MODU6N
011 100
6R
6N
6N
011 101
6R
6N
6N
011 110
6R
6N
6N
011 111
6R
6N
6N
PC-relative family encodings: Table A.24 and Table A.25 present the PC-relative family of instruction encodings. Table A.24 in traditional form, Table A.25 in the bitstring form that clearly shows the immediate varying from 19 bits to 16 bits.
bits 18-16 bits 20-19
rs
Table A.24 Release 6 PC-relative family encoding
111011.rs.TTTTT.immediate
0
1
2
00
01
10
0
000
1
001
2
010
3
011
ADDIUP6N immediate
LWP6N immediate
6N
4
100
5
101
6
110
7
111
3
11
6N
reserved (RI)
AUIP6N immediate
ALUIP6N immediate
462
The MIPS32® Instruction Set Manual, Revision 6.06

Table A.25 Release 6 PC-relative family encoding bitstrings
111011.rs.*
A.4 Release 6 Instruction Encodings
encoding
instruction
111011.rs.00.<-----immediate>
ADDIUPC6N
111011.rs.01.<----off19>
LWPC6N
111011.rs.10.<----off19>
6N
111011.rs.110.<---off18>
6N
111011.rs.1110.<---imm17>
reserved, signal RI6N
111011.rs.11110.<--immediate>
AUIPC6N
111011.rs.11111.<--immediate>
ALUIPC6N
B*C compact branch and jump encodings: In several cases Release 6 uses much tighter instruction encodings than previous releases of the MIPS architecture, reducing redundancy, to allow more instructions to be encoded. Instead of purely looking at bitfields, Release 6 defines encodings that compare different bitfields: e.g. the encoding
010110.rs rt.offset16 is BGEC if neither rs nor rt are 00000 and rs is not equal to rt, but is BGEZC if rs is the same as rt, and is BLEZC if rs is 00000 and rt is not. (The encoding with rt 00000 and arbitrary rs is the pre-Release 6 instruc- tion BLEZL.rs.00000.offset16, a branch likely instruction which is removed by Release 6, and whose encoding is required to signal the Reserved Instruction exception. )
This tight instruction encoding motivates the bitstring and constraints notation for Release 6 instruction encodings
BLEZC rt
BGEZC rt
BGEC rs,rt
BLEZL rt
010110.00000.rt.offset16,
010110.rs=rt.rt.offset16, rs!=0,
10110.rs.rt.offset16, rs!=0,
010110.00000.rt.offset16,
rt!=0
rt!=0, rs=rt
rt!=0, rs!=rt
rs=0
and the equivalent constraints indicated in the instruction encoding diagrams for the instruction descriptions in Vol- ume II. Table A.26 below shows the B*C compact branch encodings, which use constraints such as RS = RT. pre- Release 6 encodings that are removed by Release 6 are shaded darkly, while the remaining redundant encodings are shaded lightly or stippled.
Note: Pre-Release 6 instructions BLEZL, BGTZL, BLEZ, and BGTZ do not conflict with the new Release 6 instruc- tions they are tightly packed with in the encoding tables, but the ADDI, DADDI, LWC2, SWC2, LDC2 and SDC2 truly conflict.
The MIPS32® Instruction Set Manual, Revision 6.06 463

Instruction Bit Encodings
Table A.26 B*C compact branch encodings
Constraints involving rs and rt fields
rs/rt0/NZ
NZrs =/ NZrt
Constraints involving rs and rt fields
rs/rt0/NZ
NZrs =/ NZrt
0rs 0rt
useless BLEZL6R
BGEZC6N
=
0rs NZrt
BLEZC6N
BGEC6N (BLEC)
< >
NZrs 0rt
BLEZL6R
0rs 0rt
useless BLEZ
BGEZALC6N
=
0rs NZrt
BLEZALC6N
BGEUC6N (BLEUC)
< >
NZrs 0rt
BLEZ
0rs 0rt
useless BGTZL6R
BLTZC6N
=
0rs NZrt
BGTZC6N
BLTC6N (BGTC)
< >
NZrs 0rt
BGTZL6R
0rs 0rt
useless BGTZ
BLTZALC6N
=
0rs NZrt
BGTZALC6N
BLTUC6N (BGTUC)
< >
NZrs 0rt
BGTZ
0rs NZrt
0rs 0rt
BEQZALC6N
ADDI
BOVC6N
= NZrs 0rt >
rsNZ rt0,NZ
BEQC6N
< 0rs NZrt 0rs 0rt BNEZALC6N DADDI6R BNVC6N = NZrs 0rt >
rsNZ rt0,NZ
BNEC6N
< LDC26R 0rs NZrt 0rs 0rt < = NZrs 0rt >
NZrs 0/NZrt
JIC6N rt+off16
BEQZC6N rsNZ, off21
SDC26R
0rs NZrt 0rs 0rt
< = NZrs 0rt >
NZrs 0/NZrt
JIALC6N rt+off16
BNEZC6N rsNZ, off21
LWC26R
BC6N off26<<2 BALC6N off26<<2 0/NZrs 0/NZrt SWC26R 0/NZrs 0/NZrt 464 The MIPS32® Instruction Set Manual, Revision 6.06 110 010 110 110 0rs 0/NZrt 001 000 010 111 010 110 Primary Opcode 111 010 110 110 0rs 0/NZrt 011 000 rsNZ  rtNZ 000 111 rsNZ  rtNZ 000 110 Primary Opcode rsNZ  rtNZ rsNZ  rtNZ The MIPS32® Instruction Set Manual, Revision 6.06 465 A.4 Release 6 Instruction Encodings Appendix B Revision History Revision 0.90 0.91 0.92 0.95 1.00 Date November 1, 2000 November 15, 2000 December 15, 2000 March 12, 2001 August 29, 2002 Description Internal review copy of reorganized and updated architecture documentation. Internal review copy of reorganized and updated architecture documentation. Changes in this revision: • Correct sign in description of MSUBU. • Update JR and JALR instructions to reflect the changes required by MIPS16. Update for second external review release Update based on all review feedback: • Add missing optional select field syntax in mtc0/mfc0 instruction descriptions. • Correct the PREF instruction description to acknowledge that the PrepareForStore function does, in fact, modify architectural state. • To provide additional flexibility for Coprocessor 2 implementations, extend the sel field for DMFC0, DMTC0, MFC0, and MTC0 to be 8 bits. • Update the PREF instruction to note that it may not update the state of a locked cache line. • Remove obviously incorrect documentation in DIV and DIVU with regard to putting smaller numbers in register rt. • Fix the description for MFC2 to reflect data movement from the coprocessor 2 regis- ter to the GPR, rather than the other way around. • Correct the pseudo code for LDC1, LDC2, SDC1, and SDC2 for a MIPS32 imple- mentation to show the required word swapping. • Indicate that the operation of the CACHE instruction is UNPREDICTABLE if the cache line containing the instruction is the target of an invalidate or writeback invali- date. • Indicate that an Index Load Tag or Index Store Tag operation of the CACHE instruc- tion must not cause a cache error exception. • Make the entire right half of the MFC2, MTC2, CFC2, CTC2, DMFC2, and DMTC2 instructions implementation dependent, thereby acknowledging that these fields can be used in any way by a Coprocessor 2 implementation. • Clean up the definitions of LL, SC, LLD, and SCD. • Add a warning that software should not use non-zero values of the stype field of the SYNC instruction. • Update the compatibility and subsetting rules to capture the current requirements. The MIPS32® Instruction Set Manual, Revision 6.06 466 Revision 1.90 Date September 1, 2002 Description Merge the MIPS Architecture Release 2 changes in for the first release of a Release 2 processor. Changes in this revision include: • All new Release 2 instructions have been included: DI, EHB, EI, EXT, INS, JALR.HB, JR.HB, MFHC1, MFHC2, MTHC1, MTHC2, RDHWR, RDPGPR, ROTR, ROTRV, SEB, SEH, SYNCI, WRPGPR, WSBH. • The following instruction definitions changed to reflect Release 2 of the Architecture: DERET, ERET, JAL, JALR, JR, SRL, SRLV • With support for 64-bit FPUs on 32-bit CPUs in Release 2, all floating point instruc- tions that were previously implemented by MIPS64 processors have been modified to reflect support on either MIPS32 or MIPS64 processors in Release 2. • All pseudo-code functions have been updated, and the Are64BitFPOperationsEnabled function was added. • Update the instruction encoding tables for Release 2. Continue with updates to merge Release 2 changes into the document. Changes in this revision include: • Correct the target GPR (from rd to rt) in the SLTI and SLTIU instructions. This appears to be a day-one bug. • Correct CPR number, and missing data movement in the pseudocode for the MTC0 instruction. • Add note to indicate that the CACHE instruction does not take Address Error Excep- tions due to mis-aligned effective addresses. • Update SRL, ROTR, SRLV, ROTRV, DSRL, DROTR, DSRLV, DROTRV, DSRL32, and DROTR32 instructions to reflect a 1-bit, rather than a 4-bit decode of shift vs. rotate function. • Add programming note to the PrepareForStore PREF hint to indicate that it cannot be used alone to create a bzero-like operation. • Add note to the PREF and PREFX instruction indicating that they may cause Bus Error and Cache Error exceptions, although this is typically limited to systems with high-reliability requirements. • Update the SYNCI instruction to indicate that it should not modify the state of a locked cache line. • Establish specific rules for when multiple TLB matches can be reported (on writes only). This makes software handling easier. Changes in this revision: • Correct figure label in LWR instruction (it was incorrectly specified as LWL). • Update all files to FrameMaker 7.1. • Include support for implementation-dependent hardware registers via RDHWR. • Indicate that it is implementation-dependent whether prefetch instructions cause EJTAG data breakpoint exceptions on an address match, and suggest that the pre- ferred implementation is not to cause an exception. • Correct the MIPS32 pseudocode for the LDC1, LDXC1, LUXC1, SDC1, SDXC1, and SUXC1 instructions to reflect the Release 2 ability to have a 64-bit FPU on a 32- bit CPU. The correction simplifies the code by using the ValueFPR and StoreFPR functions, which correctly implement the Release 2 access to the FPRs. • Add an explicit recommendation that all cache operations that require an index be done by converting the index to a kseg0 address before performing the cache opera- tion. • Expand on restrictions on the PREF instruction in cases where the effective address has an uncached coherency attribute. • 2.00 June 9, 2003 2.50 July 1, 2005 The MIPS32® Instruction Set Manual, Revision 6.06 467 Revision History Revision 2.60 2.61 2.62 3.00 3.01 3.02 3.50 3.51 5.00 5.01 5.02 Date June 25, 2008 July 10, 2008 January 2, 2009 March 25, 2010 June 01, 2010 March 21, 2011 September 20, 2012 October 20, 2012 December 14, 2012 December 15, 2012 April 22, 2013 Description Changes in this revision: • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • Applied the new B0.01 template. Update RDHWR description with the UserLocal register. added PAUSE instruction Ordering SYNCs CMP behavior of CACHE, PREF*, SYNCI CVT.S.PL, CVT.S.PU are non-arithmetic (no exceptions) *MADD.fmt & *MSUB fmt are non-fused. various typos fixed Revision History file was incorrectly copied from Volume III. Removed index conditional text from PAUSE instruction description. SYNC instruction - added additional format “SYNC stype” LWC1, LWXC1 - added statement that upper word in 64bit registers are UNDE- FINED. CVT.S.PL and CVT.S.PU descriptions were still incorrectly listing IEEE exceptions. Typo in CFC1 Description. CCRes is accessed through $3 for RDHWR, not $4. JALX instruction description added. Sub-setting rules updated for JALX. Copyright page updated. User mode instructions not allowed to produce UNDEFINED results, only UNPRE- DICTABLE results. RECIP, RSQRT instructions do not require 64-bit FPU. MADD/MSUB/NMADD/NMSUB pseudo-code was incorrect for PS format check. Added EVA load/store instructions: LBE, LBUE, LHE, LHUE, LWE, SBE, SHE, SWE, CACHEE, PREFE, LLE, SCE, LWLE, LWRE, SWLE, SWRE. TLBWI - can be used to invalidate the VPN2 field of a TLB entry. FCSR.MAC2008 bit affects intermediate rounding in MADD.fmt, MSUB fmt, NMADD.fmt and NMSUB.fmt. FCSR.ABS2008 bit defines whether ABS fmt and NEG.fmt are arithmetic or not (how they deal with QNAN inputs). CACHE and SYNCI ignore RI and XI exceptions. CVT, CEIL, FLOOR, ROUND, TRUNC to integer can’t generate FP-Overflow exception. R5 changes: DSP and MT ASEs -> Modules
NMADD.fmt, NMSUB fmt – for IEEE2008 negate portion is arithmetic.
No technical content changes: Update logos on Cover. Update copyright page.
Fix: Figure 2.26 Are64BitFPOperationsEnabled Pseudcode Function – “Enabled” was missing.
R5 change retroactive to R3: removed FCSR.MCA2008 bit: no architectural support for fused multiply add with no intermediate rounding. Applies to MADD fmt, MSUB fmt, NMADD fmt, NMSUB fmt.
Clarification: references to “16 FP registers mode” changed to “the FR=0 32-bit reg- ister model”; specifically, paired single (PS) instructions and long (L) format instruc- tions have UNPREDICTABLE results if FR=0, as well as LUXC1and SUXC1. Clarification: C.cond fmt instruction page: cond bits 2..1 specify the comparison, cond bit 0 specifies ordered versus unordered, while cond bit 3 specifies signaling versus non-signaling.
R5 change: UFR (User mode FR change): CFC1, CTC1 changes.
468
The MIPS32® Instruction Set Manual, Revision 6.06

Revision
5.03
Date
August 21, 2013
Description
• Resolved inconsistencies with regards to the availability of instructions in MIPS32r2: MADD fmt family (MADD.S, MADD.D, NMADD.S, NMADD.D, MSUB.S, MSUB.D, NMSUB,S, NMSUB.D), RECIP.fmt family (RECIP.S, RECIP.D, RSQRT.S, RSQRT.D), and indexed FP loads and stores (LWXC1, LDXC1, SWXC1, SDXC1). The appendix section A.2 “Instruction Bit Encoding Tables”, shared between Volume I and Volume II of the ARM, was updated, in particular the new upright delta  mark is added to Table A.2 “Symbols Used in the Instruction Encod- ing Tables”, replacing the inverse delta marking  for these instructions. Similar updates made to microMIPS’s corresponding sections. Instruction set descriptions and pseudocode in Volume II, Basic Instruction Set Architecture, updated. These instructions are required in MIPS32r2 if an FPU is implemented. .
• Misaligned memory access support for MSA: see Volume II, Appendix B “Mis- aligned Memory Accesses”.
• Has2008 is required as of release 5 – Table 5.4, “FIR Register Descriptions”.
• ABS2008 and NAN2008 fields of Table 5.7 “FCSR RegisterField Descriptions” were optional in release 3 and could be R/W, but as of release 5 are required, read-only, and
preset by hardware.
• FPU FCSR.FS Flush Subnormals / Flush to Zero behavior is made consistent with
MSA behavior, in MSACSR.FS: Table 5.7, “FCSR Register Field Descriptions”,
updated. New section 5.8.1.4 “Alternate Flush to Zero Underflow Handling”.
• Volume I, Section 2.2 “Compliance ad Subsetting” noted that the L format is required
in MIPS FPUs, to be consistent with Table 5.4 “FIR Register Field Definitions” .
• Noted that UFR and UNFR can only be written with the value 0 from GPR[0]. See
section 5.6.5 “User accessible FPU Register model control (UFR, CP1 Con- trol Register 1)” and section 5.6.5 “User accessible Negated FPU Register model control (UNFR, CP1 Control Register 4)”
LLSC Related Changes
• Added ERETNC. New.
• Modified SC handling: refined, added, and elaborated cases where SC can fail or was
UNPREDICTABLE.
XPA Related Changes
• Added MTHC0, MFHC0 to access extensions. All new.
• Modified MTC0 for MIPS32 to zero out the extended bits which are writable. This is
to support compatibility of XPA hardware with non XPA software. In pseudo-code,
added registers that are impacted.
• MTHC0 and MFHC0 – Added RI conditions.
• Feature complete R6U draft of Volume II new instructions.
• Split MAX fmt-family, instruction description that described multiple instructions, into separate instruction description pages MAX fmt, MAX_A fmt, MIN fmt, MIN_A fmt.
• Mnemonic change: AUIPA changed to ALUIPC, Aligned Add Upper Immediate to PC. Now all Release 6 new PC relative instructions end in “P”.
• Renamed CMP.cond fmt -> CMP.condn.fmt, i.e. renamed 5-bit cond field “condn” to distinguish it from old 4-bit cond field.
• Cleaning up descriptions of NAL and BAL to reduce confusion about deprecation versus removal of BLTZAL and BGEZAL.
• DAHI and DATI use rs src/dest register, not rt.
• Table showing that the compact branches are complete, reversing rs and rt for BLEC,
BGTC, BLEUC, BGTUC
• Forbidden slot RI required; takes exception like delay slot; boilerplate consistency
automated.
• MOD instruction family: remainder has same sign as dividend • Updated to R6U 1.03
5.04
6.00 – R6U draft
December 11, 2013
Dec. 19, 2013 Jan 14-16, 2014
The MIPS32® Instruction Set Manual, Revision 6.06 469

Revision History
Revision
R6U draft
R6U-pre- release draft
Date
Jan 17, 2014
Jan 20, 2014
Feb 10, 2014
Feb. 11, 2014
Description
• NAL, BAL: improved confusing explanation of how NAL and BAL used to be special cases of BLEZAL, etc., instructions removed by Release 6
• Forbidden slot boilerplate: requires Reserved Instruction exception for control instructions, even if interrupted: exception state (EPC, etc.) points to branch, not for- bidden slot, like delay slot.
• Fixed bugs and changed instruction encodings: BEQZALC, BNEZALC, BGEUC, BLTUC, BLEZLC family, BC1EQZ, BC2EQZ, BC1NEZ, BC2NEZ, BITSWAP
• AUI, BAL
• Refactored “Compatibility and Subsetting” sections of Volumes I and II for reuse without replication.
• Updated Volume II tables of instructions by categories (preceding section entitled Alphabetical List of Instructions) for R6U changes.
Technical Publications preparing for release.
Summary of all R6U drafts up to this date – R6U version 1.03
• MIPS3D removed from the Release 6 architecture.
• Some 3-source instructions (conditional moves) replaced with new 2-source instruc-
tions: MOVZ/MOVN fmt replaced by SELEQZ/SELNEZ.fmt; MOVZ/MOVN
replaced by SELEQZ/SELNEZ.
• PREF/PREFE: Unsound prefetch hints downgraded; optional implementation depen-
dent prefetch hints expanded.
Free up Opcode Space
• Change encodings of LL/SC/LLD/SCD/PREF/CACHE, reducing offset from 16 bits
to 9 bits
• SPECIAL2 encodings changed: CLO/CLZ/DCLO/DCLZ
• Other changes mentioned below: traps with immediate operands removed (ADDI/
DADDI, TGEI/TGEIU/TLTI/TLTIU/TEQI/TNEI)
• Free 15 major opcodes: COP1X, SPECIAL2, LWL/LWR, SWL/SWR, LDL/LDR,
SDL/SDR, LL/SC, LLD/SCD, PREF, CACHE, as described below, by changing encodings.
Integer Multiply and Divide
470
The MIPS32® Instruction Set Manual, Revision 6.06


Integer accumulators (HI/LO) removed from base Release 6, moved to DSPr6, allowed only with microMIPS: MFHI, MTHIO, MFLO, MTLO, MADD, MADDU, MUL, MSUB, MSUBU removed.
Release 6 adds multiply and divide instructions that write to same-width register: MULT replaced by MUL/MUH; MULTU replaced by MULU/MUHU; DIV replaced by DIV/MOD; DIVU replaced by DIVU/MODU; similarly for 64-bit DMUH, etc.

Revision Date
Description
Control Transfer Instructions (CTIs)
• Branch likely instructions removed by Release 6: BEQL, etc.
• Enhanced compact branches and jumps provided
• No delay slots; back-to-back branches disallowed (forbidden slot)
• More complete set of conditions: BEQC/BNEC, all signed and unsigned reg-reg com-
parisons, e.g. BLTC, BLTUC; all comparisons against zero, e.g. BLTZC
• More complete set of conditional procedure call instructions: BEQZALC, BNE-
ZALC
• Large offset PC-relative branches: BC/BALC 26-bit offset (scaled by 4); BEQZC/
BNEZC 21-bit offset
• JIC/JIALC: “indexed” jumps, jump to register + sign extended 16-bit offset
• Trap-in-overflow adds with immediate removed by MIOPSr6: ADDI, DADDI;
replaced by branches on overflow BOVC/BNVC.
• Redundant JR.HB removed, aliased to JALR.HB with rdest=0.
• BLTZAL/BGEZAL removed; not used because unconditionally wrote link register SSNOP identical to NOP.
Misaligned Memory Accesses
• Unaligned load/store instructions (LWL/LWR, etc.) removed from Release 6. Support
for misaligned memory accesses must be provided by a Release 6 system for all ordi-
nary loads and stores, by hardware or by software trap-and-emulate.
• CPU scalar ALIGN instruction
Address Generation and Constant Building
• Instructions to build large constants (such as address constants): AUI (Add upper
immediate), DAHI, DATI.
• Instructions for PC-relative address formation: ADDIUPC, ALUIPC.
• PC-relative loads: LWP, LWUP, LDP.
• Indexed FPU memory accesses removed: LWXC1, LUXC1, PFX, etc.
• Load-scaled-address instructions: LSA, DLSA
• 32-bit address wrapping improved.
DSP ASE
• DSP ASE and SmartMIPS disallowed; recommend MSA instead
• DSPr6 to be defined, used with microMIPS.
• Instructions promoted from DSP ASE to Base ISA: BALIGN becomes Release 6
ALIGN, BITREV becomes Release 6 BITSWAP
The MIPS32® Instruction Set Manual, Revision 6.06 471

Revision History
Revision
Date
Description
FPU and co-processor
• Instruction encodings changed: COP2 loads/stores, cache/prefetch, SPECIAL2:
LWC2/SWC2, LDC2/SWC2
• FR=0 not allowed, FR=1 required.
• Compatibility and Subsetting section amended to allow a single precision only FPU
(FIR.S=FIR.W=1, FIR.D=FIR.L=0.)
• Paired Single (PS) removed from the Release 6 architecture, including: COP1.PS,
COP1X.PS, BC1ANY2, BC1ANY4, CVT.PS.S, CVT.PS.W.
• FPU scalar counterparts to MSA instructions: RINT fmt, CLASS fmt, MAX/MAXA/
MIN/MINA fmt.
• Unfused multiply adds removed: MADD/MSUB/NMADD/NMSUB fmt
• IEEE2008 Fused multiply adds added: MADDF/MSUBF fmt
• Floating point condition codes and related instructions removed: C.cond fmt
removed, BC1T/BC1F, MOVF/MOVT.
• MOVF/MOVT fmt replaced by SEL fmt
• New FP compare instruction CMP.cond fmt places result in FPR and related
BC1EQZ/BC2EQZ
• New FP comparisons: CMP.cond fmt with cond = OR (ordered), UNE (Unordered
or Not Equal), NE (Not Equal).
• Coprocessor 2 condition codes removed: BC2F/BC2T removed, replaced by
BC2NEQZ/BC2EQZ
Recent R6U architecture changes not fully reflected in this draft:
• This draft does not completely reflect the new 32-bit address wrapping proposal but still refers in some places to the old IAM (Implicit Address Mode) proposal.
• This draft does not yet reflect constraints on endianness, in particular in the section ion Misaligned memory access support: e.g. code and data must have the same endi- anness, Status.RE is removed, etc.
• BC1EQZ/BC1NEZ will test only bit 0 of the condition register, not all bits.
• This draft does not yet say that writing to a 32-bit FPR renders upper bits of a 64 bit
FPR or 128 bit floating point register UNPREDICTABLE; it describes the old pro- posal of zeroing the upper bits.
Known issues:
472
The MIPS32® Instruction Set Manual, Revision 6.06

• •


This draft describes Release 6, as well as earlier releases of the MIPS architecture. E.g. instructions that were present in MIPSr5 but which were removed in Release 6 are still in the manual, although they should be clearly marked “removed by Release 6” to indicate that they have been removed by Release 6.
R6U new instruction pseudocode is 64-bit, rather than 32-bit, albeit attempting to use notations that apply to both.
Certain new instruction descriptions are “unsplit”, describing families of instructions such as all compact branches, rather than separate descriptions of each instruction. This facilitates comparison and consistency, but currently allows certain MIPS64 Release 6 instructions to appear inappropriately in the MIPS32 Release 6 manual. A future release of the manual will “split” these instruction family descriptions, e.g. the compact branch family will be split up into at least 12 different instruction descrip- tions.
R6U requires misalignment support for all ordinary memory reference instructions, but the pseudocode does not yet reflect this. Boilerplate has been added to all existing instructions saying this.
The new R6U PC-relative loads (LWP, LWUP, LDP) in this draft incorrectly say that misaligned accesses are permitted.

Revision
R6U-pre- release draft
Date
Feb. 13, 2014
Description
• ALIGN/DALIGN: clarified bp=0 behavior
• ALIGN/DALIGN pseudocode used || as logical OR rather than MIPS’ pseudocode
concatenate.
• Removed incorrect note about not using r31 as a source register to BAL.
• Release 6 requires BC1EQZ/BC1NEZ if an FPU is present, i.e. they cannot signal RI. • R6U 1.05 change: BC1EQZ/BC1NEZ test only bit 0 of the FPY; changed from test-
ing if any bit nonzero; helps with trap-and-emulate of DP on an SP-only FPU.
• Known problem: R6U 1.05 change not yet made: all 32-bit FP operations leave upper bits of 64 bit FOR and./or 128-bit MSR unpredictable; helps with trap-and-emulate of
DP on an SP-only FPU.
• Clearly marked all .PS instructions as removed via removed by Release 6 in instruc-
tion format.
• DMUL, DMULTU, DDIV, DDIVU marked removed by Release 6
• Started using =Release 6 notation to indicate that an instruction has been changed but
is still present. JR.HB =Release 6, aliased to JALR.HB. SSNOP =Release 6, treated
as NOP.
• Noted that BLTZAL and BGEZAL are removed by Release 6, the special cases
NAL=BLTZAL with rs=0 and BAL=BGEZAL with rs=0, remain supported by
Release 6.
• Marked conditional traps with immediate removed by Release 6.
• Overeager propagation of r31 restriction to non-call instructions5 removed.
• Emphasized that unconditional compact CTIs have neither delay slot nor forbidden
slot.
• SDBBP updated for R6P facility to disable if no hardware debug trap handler
• UFR/UNFR (User-mode FR facility) disallowed in Release 6: changes to CTC1 and
CFC1 instructions.
• Last minute change: BC1EQZ.fmt and BC1NEZ fmt test only bit 0, least significant bit, of FPR.
Known issues:
• Similar changes to SEL fmt, SELEQZ fmt, SELNEZ fmt not yet made.
• FPU truth consuming instructions (BC1EQZ fmt, BC1NEZ.fmt, SEL fmt, SELEQZ fmt, SELNEZ fmt) change completed: test bit 0, least-significant-bit, of FPR containing condition.
• Production Release.
• Add DVP and EVP instructions for multithreading.
• Add POP and SOP encoding nomenclature to opcode tables in appendix A
• JIC format changed from JIC offset(rt) to JIC rt, offset.
• JIALC format changed from JIALC offset(rt) to JIALC rt, offset. • ‘offset’ removed from NAL format.
• Fixed many inconsistencies; no functional impact. • RDHWR updates for Release 6.
• WAIT updates for Release 6.
• CFC1/CTC1 UFR-related text reworded.
• CFC1/CTC1 FRE-related text added.
• Added LLX/SCX(32/64) instructions.
• Jump Register ISA Mode switching text reworded.
• MisalignedSupport() language in ld/st pseudo-code reworded.
• Release 6 behaviour added to move-to/from instructions: return 0,nop.
• TLBINV/TLBINVF description and pseudocode corrected and clarified. • ALIGN/DALIGN pseudocode cleaned up; removed redundancy.
• Removed “Special Considerations” section from Bc
• Language clarified in PREF/PREFE tables; no functional change.
R6U ARM Volume II 6.00 pre- liminary release
post-6.00
6.01
6.02
6.03
February 14, 2014
February 20, 2014
December 1, 2014
December 10, 2014
September 4, 2015
The MIPS32® Instruction Set Manual, Revision 6.06 473

Revision History
474
The MIPS32® Instruction Set Manual, Revision 6.06
Revision
6.04
Date
November 13, 2015
Description
6.05
June 3, 2016
• J/JAL now indicated as deprecated (but not removed).
• DVP: Added text indicating that a disabled VP will not be re-enabled for execution on
deferred exception.
• CACHE/CACHEE: Undefined operations are really NOP.
• CMP.condn.fmt: removed fmt related text in description section. .S/.D explicitly
encoded.
• Fixed minor textual typos in MAXA/MINA fmt functions.
• DERET: Restriction – if executed out of debug mode, then RI, not UNDEFINED.
• TLBWR: Updated reference to Random. No longer supported in Release 6.
• PCREL instructions: Added PCREL minor opcode table, fixed conditional text bugs
in register reference.
• BC1F/BC1FL/BC1T/BC1TL: Removed last paragraph of historical information sec-
tion. These instructions can be immediately preceeded by instruction that sets cond.
code.
• JIALC: Restructured operation section using ‘temp’ to avoid false hazard of link
update overwriting source.
• LUI: Fixed conditional text errors related to the encoding table. microMIPS appeared
in MIPS.
• JIALC/JIC: Updated to indicate effect on ‘ISAMode’.
• Fixed typo ROUND/TRUNC/FLOOR/CEIL.W fmt. Range value should be 231-1 not
263-1.
MIPS64 only:
• DMFC0/DMTC0: Now indicates what happens with 32-bit COP0 registers.
MIPS32 and MIPS64:
• RDHWR: Fixed typo in the RDHWR register number table header; rs changed to rd. Changed Double-Width LLX/SCX to Paired LL/SC.
• DMTC2: Changed CPR[2, rd, sel] to CP2CPR[Impl].
• WAIT: Fixed a bit range typo.
• LSA: Removed the word optional; the scaling shift on rs is not optional.
• CACHE: Fixed typo; CACHE has a 9-bit offset.
• SYSCALL, TEQ, TGE, and TGEU: If COP0 BadInstr is implemented, the code field
may be obtained from BadInstr.
• JALR, JALR.HB, JIALC, JIC, JR, and JR.HB: Updated condition for
PC  temp in the Operation pseudocode. MIPS32:
• Removed the LLX, LLXE, SCX, and SCXE instructions.
• Added the LLWP, LLWPE, SCWP, and SCWPE instructions. MIPS64:
MIPS32 and MIPS64:
• •
Removed the LLDX and SCDX instructions.
Added the LLDP, LLWP, LLWPE, SCDP, SCWP, and SCWPE instructions.

Revision Date
6.06 December 15, 2016
Description
MIPS32 and MIPS64:
• Added CRC32B, CRC32H, CRC32W, CRC32CB, CRC32CH, CRC32CW.
• DVP/EVP instructions incorrectly used ‘rs’. Changed to use ‘rt.’
• Added GINVI, and GINVT instructions.
• SC, SCE, SCWP, SCWPE: Updated description for uncached handling.
• MTHC0: updated description, fixed typo. ‘COP2’ changed to ‘COP0’.
• MTC0: changed Config5MVH to new Config5XPA.
• DERET: updated pseudocode to describe what happens if DebugDM=0.
• TLT, TLTU, TNE: Mention that contents of the code field can be retrieved from
COP0 BadInstr if present.
• Added ArchitectureRevision(), IsCoprocessorRegisterImplemented(), and IsCopro-
cessorRegisterExtended() pseudocode descriptions (ARM only, not AFP)
MIPS64:
• Added CRC32D, CRC32CD.
• SCD, SCDP: Updated description for uncached handling.
• LLDP: Fixed typos. Swapped rd and rt in the GPR references and description; in the
pseudocode, doubled the bit range in both cases where GPR is loaded. (typos).
• SCWP, SCWPE: fixed typo due to conditional text.
The MIPS32® Instruction Set Manual, Revision 6.06 475
Copyright © Wave Computing, Inc. All rights reserved. www.wavecomp.ai